Purpose: Children who have late language emergence, or are late talkers (LTs), have substantially lower vocabulary levels than their peers, on average. Notably, differences in how researchers define "who" comprises LTs can lead to inconsistencies across findings. The current study examined how the number of children identified as LTs differs when using different parameters for identification in a low-socioeconomic status (SES) sample. Method: Low-SES mothers (n = 238) completed the MacArthur-Bates Communicative Development Inventories: Words and Sentences (CDI:WS) as part of a larger longitudinal study. Using percentile scores, children were identified as LTs or non-LTs using different sets of parameters. Descriptive and chi-square analyses were used to examine how the different parameters changed the percentages of children identified as LTs. Results: Depending on the parameters used, the prevalence of LTs in our low-SES sample ranged from 28% to 43%, which is higher than the prevalence in population-based studies (10%-20%). Using 3rd edition norms by sex for children ages 24-30 months and using a 10th percentile cutoff value, the prevalence of LTs was 29%. Conclusions: Reporting and agreeing upon the most robust parameters for LT identification is critical both for (a) research replicability and comparison across studies and (b) researchers and practitioners to accurately identify LTs and provide the appropriate support to them and their families.