Classroom climate presents dynamic relationships within the classroom and indicates the way the child perceives educational, psychological, social and physical aspects of classroom environment. The greatest part of previous research on classroom climate was implemented with the goal of examining the relationship between students' perceptions of the climate and cognitive and affective aspects of learning. It has established that student achievements are greater in the classroom climates that are well organised, goal-directed, less tense and more connected with students. Self-efficacy is defined as an individual's belief he/she is capable of doing certain actions needed for achieving a specific goal. Social-cognitive theory, within whose frame the self-efficacy construct emerged, understands human functioning as a result of dynamic interaction of personal, behavioral and environmental influences. It is assumed that self-efficacy influences behaviour and environment, and vice versa. Student engagement is defined as a measure of inclusion, connection and commitment of the students to academic and social activities in school. The three-component model of engagement comprises cognitive, behavioral and emotional component of engagement. Cognitive engagement describes the extent to which the students perceive the importance of instruction and their attitudes toward learning. Behavioral engagement entails student presence and investing effort in classes. Emotional engagement encompasses student sense of belonging to a school and school-related emotions such as happiness, excitement and enjoyment. School satisfaction is defined as the cognitive-affective evaluation of school satisfaction. It is well-known that students' school experiences influence their cognitive development, physical and mental health, and their later possibilities and choices of further education. Supporting school climate in which teachers care about student satisfaction presents the basis for developing acceptable forms of behaviour, overall health and achieves satisfaction with student life. The fundamental idea of this research ensues from the presumption that human life is highly co-dependent: what an individual does effects the welfare of his/her surroundings and vice versa. This work assumes that the classroom climate is linked to student school satisfaction, but with the mediating role of self-efficacy and engagement. The goal of this research was to examine the way students perceive the classroom climate, the extent of their self-efficacy, engagement and school satisfaction. Besides, the research explored the existence of differences in perceptions with regards to gender, age and academic achievement in Croatian language, mathematics and English language. It also examined the contribution of classroom climate to the explanation of self-efficacy, student engagement and school satisfaction. Structural equation modelling (SEM) was used to establish whether student self-efficacy and engagement function as an indirect path, i.e. mediator variables in the relationship between the classroom climate and students' school satisfaction. Methodology The research involved 685 students of two primary schools in Zagreb. The participation in the research was anonymous and voluntary, with the previously obtained parent consent. After excluding the outliers, 597 subjects were included in the final data processing. The research utilised four questionnaires: My Class Inventory (Fisher & Fraser, 1981), Self-efficacy for self-regulated learning (Zimmerman et al., 1992), The Behavioral-Emotional-Cognitive School Engagement Scale (Li & Lerner, 2013), and for school satisfaction one dimension of the questionnaire Multidimensional Students' Life Satisfaction Scale, MSLSS (Huebner, 1994; Huebner et al., 1998). The factor analysis of My Class Inventory questionnaire extracted five factors: cohesiveness, satisfaction, competitiveness, friction and difficulty. The reliability coefficients are in range from α = ,783 to α = ,843. Due to low reliability for dimension difficulty (α = ,589), it is left out of further analyses. The obtained factor solution explains 54,9% of the variance. The factor analysis of questionnaire Self-efficacy for self-regulated learning yielded one factor which explains 44,47% of the variance in accord with the original solution (α = , 868). The factor analysis of the questionnaire on school satisfaction explains 50,15% of the variance (α = ,851). The factor analisys of The Behavioral-Emotional-Cognitive School Engagement Scale questionnaire extracted three factors that together explain 53,3% of the engagement variance. The obtained reliability coefficients are: α = ,779 for emotional engagement, α = ,707 for behavioral engagement, and α = ,699 for cognitive engagement. Besides the applied questionnaires, data was collected on the students' demographic characteristics - gender, form and final marks in Croatian, mathematics and English. Results and Discussion After composite results for each subscale were calculated, the fit of the data for further analyses was checked. The diagnostic evaluation of the distribution's shape and of absolute skewness and kurtosis values was implemented. After removing the extreme results, the obtained values indicate normal distribution of the results. The stated results were confirmed with the evaluation of quantile-quantile charts and histograms. The obtained values for dimensions of classroom climate show highly estimated competitiveness among students (M = 3,49; SD = 0,82) and class satisfaction (M = 3,57; SD = 0,762). Student cohesiveness is estimated as somewhat lower (M = 3,36; SD = 1,04), and the dimension of friction expectedly as the lowest (M = 2,70; SD = 0,91). It can be inferred from the presented data that the participants equally perceive both positive and negative classroom climate. If the results of class satisfaction are compared (M = 3,57; SD = 0,76) with those of school satisfaction (M = 3,47; SD = 0,75), it is noticeable that the students are slightly more satisfied with the class. Such result is expected because the class is still a smaller community they belong to, so it can be assumed they are satisfied with the way it functions. However, significantly higher results were expected in student satisfaction assessments. The obtained median for self-efficacy (M = 4,16; SD = 0,55) shows that the examined students believe in their own abilities of self-regulated learning. It indicates the students’ belief in their own competence and their feeling of being capable of independent and efficient organisation of their own work. The obtained medians for engagement dimensions point out that the participants attribute the highest values to the cognitive (M = 4,37; SD = ,503), followed by behavioral (M = 3,75; SD = ,558) and emotional engagement (M = 3,72; SD = ,755). Hence, the students evaluate the importance of schooling and their own attitudes toward learning highly, whereas they attribute somewhat lesser values to their own activity, performing school obligations and their sense of belonging to the school. One of the research tasks was to examine the differences in student perceptions with regards to form (age), gender and the achieved final grades. The research included students from the 4th to the 8th form of primary school. With regards to the specificities in the classes organisation, the sample was divided into two groups: students of class and subject teaching. Testing the significance of the differences with the use of the Student t-test has shown that students in class teaching (4th grade) differ from the students in subject teaching (5th to 8th form) with statistical difference in all the examined dimensions except competitiveness. In that, younger students evaluate all dimensions with higher values, except dimension friction. It can be assumed that the difference in the perception of positive classroom climate is the result of work organisation, because students in class teaching work with smaller number of teachers so the possibility of better cohesiveness is greater than in older students. It is possible that younger students, due to lower level of educational demands set before them, are more convinced of their own ability. Older students are faced with greater number of subjects and larger content scope to learn so they are probably less convinced of their own abilities. The results show statistically significant differences in the perception of all engagement dimensions between students in class and subject teaching, in which younger students evaluate all engagement dimensions significantly higher. The obtained differences indicate that younger students are more aware of the importance of schooling and performing school obligations and they feel better in school. The results also show that younger students are more satisfied with school than the older students. As previously assumed, younger students, in class teaching, probably receive greater attention from the teacher and participate in making class decisions more. It can be assumed that such way of cooperation and the possibility to meet the students' needs and interests contribute to their greater school satisfaction. In order to examine the differences between the girls and boys in their assessments of classroom climate, self-efficacy, engagement and school satisfaction, the Student t-test was utilised. The obtained results indicate the existence of statistically significant differences in self-efficacy and behavioral engagement. Although significant differences in assessments were not expected, it was shown that the girls are more self-efficient and behaviorally more engaged than the boys, but they are not different in the other dimensions. One of the tasks of this research was to examine the existence of differences in perceptions of classroom climate, self-efficacy, engagement and school satisfaction between the students with regards to final marks in Croatian language, mathematics and English language. After gathering all the data about the participants, it was revealed that the median of all final grades was 4,57, and even 50,25% of the examined students have A grades in all three school subjects. The participants were divided into four groups: A (N = 300; grade point average 5,00), B (N = 103; grade point average 4,67), C (N = 71; grade point average 4,33), D (N = 123, grade point average from 2,00 to 4,33). One-way variance analysis was used and statistically significant differences determined in all dimensions apart from competitiveness (classroom climate). Students with the highest grade point average are different with statistical significance from students from D group in all the examined variables except competitiveness. All groups of students attribute the highest assessments to self-efficacy and cognitive engagement dimensions, but are different in these assessments with statistical significance. If we assume that hard and continuous work is what lies behind these excellent grades, then such results only confirm good organisation and belief in one's own abilities is required for success. It was revealed that A students are most satisfied with school and perceive positive classroom climate more than the other students. It seems that the achieved student success empowers satisfaction over it, i.e. satisfaction with school in general. Pearson correlation coefficient was used to examine the connection between the dimensions of classroom climate, self-efficacy for self-regulated learning, emotional, cognitive and behavioral engagement and student school satisfaction. The obtained results from the overall sample of participants have shown the existence of statistically significant correlation between all these variables: they are in range from low, negative (r = -,10; p < 0,05) between competitiveness and cognitive engagement dimensions, to moderately high (r = ,75; p < 0,01) between emotional engagement and school satisfaction. In order to examine to which extent classroom climate contributes to the explanation of self-efficacy, engagement and school satisfaction, five hierarchical regression analyses were implemented. In the first step, demographic variables were added (gender and form) and the average grade from the three school subjects. In the second step, positive factors of classroom climate were added (cohesiveness and satisfaction), and in the last, third step, negative factors of classroom climate (competitiveness and friction) in order to establish the individual contribution, i.e. rise in regression coefficients with the each mentioned variable. The use of regression analysis found that grade (age) explains 8,4% of the self-efficacy variance, the average final grade 4,9%, and 0,3% pertains to gender, which is not a significant predictor. Classroom satisfaction, as a dimension of positive classroom climate, is a significant predictor and explains 4,7% of self-efficacy variance. The students' cohesiveness and dimensions of negative classroom climate are not significant predictors of self-efficacy. Regression analysis has shown that demographic factors, final grade and classroom climate explain 24,6% of cognitive engagement variance. In that, gender and grade are not significant predictors, while form/age explains 20,8% of the variance. Dimensions of positive classroom climate added in the second step increase the percentage of cognitive engagement variance by 2,7%, in which classroom satisfaction alone is a significant predictor. Negative classroom climate factors are not significant predictors for explaining cognitive engagement. The results of regression analysis show that demographic factors, final grade and classroom climate explain 38,2% of emotional engagement variance. In that, form/age explains 12%, and average grade 5,8% of emotional engagement. Classroom satisfaction explains 20,4%, and cohesiveness 7,7% of emotional engagement variance. Negative classroom climate factors do not contribute to explaining emotional engagement. The results of regression analysis show that demographic factors, final grade and classroom climate explain 29,2% of behavioral engagement variance. Gender was not proven as a significant predictor, whereas form/age explains 9,3%, and the average grade 17,1% of behavioral engagement. In the second step, the percentage of the explained variance has risen 1,7%, in which classroom satisfaction presents a significant predictor. Factors of negative classroom climate are not significant predictors of behavioral engagement. The results of regression analysis show that demographic factors, final grade and classroom climate explain 14,6% of school satisfaction variance. Form/age is a significant predictor and explains 13,7% of school satisfaction variance, and average grade 6,8%. In the second step, two factors of positive classroom climate were added and the percentage of the explained variance increased by 13,4%. Classroom satisfaction explains 17,7% and cohesiveness 2,7% of school satisfaction variance. Negative classroom climate factors are not significant predictors. To address the last research problem, structural equation modelling was used. The final structural model had the following latent variables and indicators: positive classroom climate (2 indicators: satisfaction and cohesiveness), negative classroom climate (2 indicators: friction and competitiveness), self-efficacy (3 indicators), engagement (3 indicators: cognitive, behavioral and emotional engagement), and school satisfaction (2 indicators). Measurement (χ2 = 149,89; χ2/df = 3,57; CFI = ,96; TLI = ,93; RMSEA = ,07) and structural (χ2 = 121,82; χ2/df = 2,4; CFI = ,97; TLI = ,96; RMSEA = ,05) model showed acceptable fit. In testing the path between classroom climate as a predictor, self-efficacy as a mediator and engagement as a criterion, the obtained values have indicated a significant contribution of classroom climate to self-efficacy ( = ,54, p < ,001), and a direct and significant, positive contribution to engagement ( = ,50, p < ,001). When it comes to negative classroom climate, there is no statistically significant contribution to self-efficacy ( = ,21, p > ,05), but its contribution to engagement is statistically significant, and, opposite to expectations, positive ( = ,28, p < ,05). Finally, the contribution of self-efficacy as a mediator to engagement is also statistically significant, positive and moderately high ( = ,79, p < ,001). All results are in accord with the expectations, except the result indicating negative classroom climate has a positive effect on student engagement. The cause of this can be that one part of friction and competiveness variance (competitiveness, more probable) has a positive correlation with the engagement aspects, besides the negative (e.g. although competitive climate is considered mostly negative and stressful for students, a certain amount of competitiveness can have a stimulating effect and with that be in correlation with increased engagement). Path analysis shows that self-efficacy realises the effect on overall school satisfaction only through engagement. Both effects are statistically significant and positive. Self-efficacy contributes to engagement with  = ,68 (p < ,001), while engagement also shows a significant positive effect on school satisfaction ( = ,66; p < ,001). The final structured model shows positive classroom climate has significant and positive effects on self-efficacy ( = ,34; p < ,001) and on engagement ( = ,58; p < ,001). However, after introducing the mediators into the analysis, the direct effect of positive classroom climate on school satisfaction has become closer to zero and statistically insignificant ( = -,01; p > ,05). It indicates the complete mediation from self-efficacy and engagement to the relationship between positive classroom climate and school satisfaction. Self-efficacy also has a significant positive effect on the explanation of engagement ( = ,74; p < ,001). At the same time, negative classroom climate has a positive contribution to engagement ( = ,37; p < ,01), which then leads to the final effect on school satisfaction ( = ,74; p < ,001). Age, i.e. students' form was controlled in the model. The model explained 56% of the variance of school satisfaction in total. The evaluation of the presupposed model with the use of structural modelling has confirmed the previous hypotheses. It was shown that the students who perceive positive classroom climate have greater self-efficacy and engagement and are satisfied with school. Positive classroom climate at the same time has a direct effect on student engagement making them more satisfied with school. Regardless of the research examining only the dimensions of students' satisfaction with the class and their cohesiveness, the results are in accord with previous research which included additional dimensions of positive classroom climate. The basic assumption that the students who feel good in their class will be more industrious and dedicated to school work, which finally leads to their school satisfaction, has proven to be correct.