Children's television programs from public and commercial networks were videotaped: PBS—Mr. Rogers's Neighborhood, Sesame Street, and Villa Alegra; Commercial—Action programs, Comedy/Adventure programs, and Commercials. Random 2–3 minute segments were transcribed and the dialogue of individual characters printed for analysis. In Analysis 1,11 trained linguistic coders assessed 48 transcripts (45–110 words) for 3 7 language variables previously found to differentiate speaker gender. A multiple discriminant analysis showed significant ability (94% accuracy) in predicting character sex from 13 linguistic variables. In Analysis 2, the transcripts were rated by 221 naive observers using the Speech Dialect Attitudinal Scale. MANOVA results indicated that, in general, the male and female characters' language led to character perceptions consistent with the Gender-Linked Language Effect: female characters rated higher on Socio-Intellectual Status and Aesthetic Quality; male characters rated higher on Dynamism. In Analysis 3, multiple regression analyses suggested a causal link between 9 of the 13 gender-discriminating variables from Analysis 1 and all three attributional dimensions in Analysis 2. Results are discussed in terms of sex-role stereotypes.