The autism spectrum disorder is defined as a neurodevelopmental disorder occurring in the first 3 years of life, manifested in persistent deficiencies in social communication and social interaction and limited, repetitive patterns of behaviour, interest and activity (DSM 5, 2013). ASHA considers it necessary to include language components as well, considering that discrepancies in language form and content are important clinical indicators of the existence of disorder (ASHA, 2012). Children with PSA do not use the same strategies of acquiring new words as children with typical development. The aim of this paper is to examine the share of certain word types in the expressive vocabulary of children with ASD whose vocabulary span is 200 to 400 words. According to English and Croatian literature, in typical development children the proportion of verbs is increasing during this phase compared to the share of nouns (Bates et al., 1994; Hržica, 2007). However, the hypothesis is that in children with PSA noun will continue to be dominant in their expressive vocabulary. The second aim is to determine whether there are differences between data collected by the Communicative Development Inventories (Kovacevic et al., 2007) and the method of analysis of language samples using types and tokens. Based on literature, it is assumed that there will be differences between the two methods. The sample consists of 5 children who meet the criteria for diagnosis of PSA according to ADOS-2 and DSM-5. Parents fulfilled the communication development scale, and language samples were collected by video recording semi-structured interaction between parents and children. The recorded voice samples were transcribed in the software package for the processing of children's language CLAN and coded according to CHAT (MacWhinney, 1996). The share of particular types of words for each method was then calculated. Individual word types in types and tokens were added separately. The results of the conducted analysis point to the obvious predominance of nouns during the processing of data obtained by the CDI, while the results of the processing of language samples indicate the uniformity of the share of nouns or verbs, i.e. the slight predominance of one or the other word type. In addition to this quantitative difference, there are certain qualitative differences between the methods used. These results are important because of insufficient research on vocabulary composition of children with PSA in the Croatian language and for designing targeted support.