Echolalia refers to the repetition of other people's utterances, phrases or words, and which occurs in all children who are developing typically prior to their acquisition of the basics of their mother tongue, i.e. between the first and second year of life, and at most up to the 30th month of life. Afterwards, it noticeably disappears from speech, as other language and communication skills start to develop. However, if echolalia persists longer than that, it can be a very important clinical marker for diagnosing autism spectrum disorder (ASD), and is also one of the most noticeable features in the population with ASD. Echolalia is often described as a dysfunctional and stereotyped behavior, although there is research showing that children with ASD use echolalia for multiple communicative functions and that it may play an important role in language development. Since echolalia can be present in early language development both in children with ASD and in typically developing children, the aim of this paper is to compare the features of echolalia in those two groups of children in stages of early language development. The differences in the frequency of using echolalia in general, as well as delayed echolalia, were analyzed, and as an additional issue, the differences in the frequency of using mitigated echolalia in children with ASD and typically developing children in early stages of language development (vocabulary size of 200-400 words) were also analyzed. 10 typically developing children and 10 children diagnosed with ASD participated in the research. The group of children with ASD was on average chronologically older due to differences in the time and rate of acquisition of language skills. Echolalia characteristics were analyzed using transcripts constructed on the basis of videotapes of semi-structured child-parent interaction. The collected data were processed with the non-parametric Mann-Whitney U test in the SPSS program. The results showed that children with ASD in early language development use echolalia more often than typically developing children, but the differences in the frequency of using delayed echolalia were negligible and not significant. Mitigated echolalia was used by children from both groups in a large proportion, and it turned out that even in this type of echolalia, the two groups did not differ significantly. However, it is interesting to see that a slightly higher proportion of mitigated echolalia was observed in children with ASD, which opens up a new, still an under-researched question.