Cellular automata are computational models based on cells that self-reproduce. According to the literature, these automata are simple-structured yet able to yield complex patterns, which implies in their adoption as pseudorandom generators. Taking that into account, the main objective of this work was to analise the statistical quality of pseudorandom sequences produced by the elementary and totalistic version of such automata. As a result, it was possible to verify that this model of computation is not adequate for pseudorandom sequence generation, as it shows a low statistical quality. These results indicates that the adoption of such pseudorandom generators is not recommend because they can harm the applications in which they are used.