The thesis estimates the probabilities of employment and gross-2 hourly wages (which is equal to total labour cost per hour) for inactive and unemployed persons in Croatia. The analysis was conducted separately for men and women. The differences in characteristics (age, education, marital status, place of residence, number of children, economic activities) of employed persons in comparison to the unemployed or inactive persons are analysed. Models for estimating gross-2 hourly wage determine how personal characteristics affect gross-2 hourly wage and probability to being employed. The microdata used in the research come from the Survey on Income and Living Conditions (Anketa o dohotku stanovništva, ADS). The survey has been conducted by Croatian Bureau of Statistics (Državni zavod za statistiku, DZS). The selected sample contains data for 6,403 households with 16,948 persons. Every person has its own sampling weight, which is calculated by DZS. These sampling weights enable the aggregation of sample data to the whole population level. All estimations and calculation in this thesis are performed using personal sampling weights. In samples like ADS, data on wages are recorded only for people who are actually working, whereas for unemployed and inactive persons no wage data are present. Therefore, if wage data on non-working population are required, they have to be somehow estimated. In this thesis, estimation is performed in the following way. Gross-2 hourly wages are estimated using multiple linear regression model. However, if the sample selection problem exists, such estimates may be biased and the use of tobit II model is recommended. In Croatian case it is obtained that the sample selection problem is present for women and the gross-2 hourly wages have to be estimated using tobit II model. For men sample selection problem does not occur. The estimated coefficients in both models are almost identical and the gross-2 hourly wages can be estimated using multiple linear regression model.