Determinants of risk behavior responses immediately following HIV diagnosis remain poorly understood. Literature reviews indicate a persistent failure to apply scientific rigor and theory to this field. Two empirical studies were conducted following a behavioral ecological model. Using interview and clinical chart data from a 2005 study of determinants of late HIV testing in Tijuana, analyses were conducted with two samples: 108 men who have sex with men (MSM), and 150 heterosexual men (n=64) and women (n=86). Potential predictors were selected from pre-diagnosis data and tested in a logistic regression model predicting sexual risk reduction after diagnosis. Nearly 40% of MSM remained at moderate to high risk for sexual transmission. The model significantly explained an estimated 28.4% (Nagelkerke R_) of the variance in sexual risk reduction from pre- to post-diagnosis, x_(9, N=106)=24.93, p<0.01. Significant independent predictors of decreased sexual risk behavior (p<0.05) after accounting for the other variables in the model were: a close relationship with a sexual partner before diagnosis (AOR 23.50 [95% CI 23.17—23.84] p=0.008), the interaction of peer risk behavior with a close relationship with a sexual partner before diagnosis (AOR 0.85 [95% CI 0.850—0.852] p=.024); and the interaction of years resided in a high HIV prevalence environment with education (AOR 1.03 [95% CI 1.029—1.030] p=.023). Some relationships identified in the model were not in the predicted direction. Nearly 75% of the heterosexual sample decreased sexual risk behavior from pre-diagnosis. The model significantly explained an estimated 26.4% (Nagelkerke R_) of the variance in sexual risk behavior reduction, x_(8, N=150)=29.81, p<0.001. Accounting for the other variables in the model, age at diagnosis and four theoretically-associated variables were significant independent predictors of decreased sexual risk behavior (p<0.05): (1) delayed testing because did not feel sick (AOR 0.228 [95% CI 0.065-0.797] p=.021); (2) had a close relationship with a sexual partner before diagnosis (AOR 0.339 [95% CI 0.129—0.890] p=0.028); (3) had sex in exchange for money, food, drugs, or housing before diagnosis (AOR 0.221 [95% CI 0.070—0.700] p=.010); and (4) believe was infected by steady partner (AOR 0.302 [95% CI 0.112—0.813] p=.018).