Background Stigma against persons with mental illness is a universal phenomenon, but culture influences the understanding of etiology of mental illness and utilization of health services. Methods We validated Kiswahili versions of three measures of stigma which were originally developed in the United Kingdom: Community Attitudes Toward the Mentally Ill Scale (CAMI), Reported and Intended Behaviors Scale (RIBS) and Mental Health Awareness Knowledge Schedule (MAKS) and evaluated their psychometric properties using a community sample (N = 616) in Kilifi, Kenya. Results Confirmatory factor analysis confirmed the one-factor solution for RIBS [root mean-squared error of approximation (RMSEA) < 0.01, comparative fit index (CFI) = 1.00, Tucker–Lewis index (TLI) = 1.01] and two-factor solution for MAKS (RMSEA = 0.04, CFI = 0.96, TLI = 0.95). A 23-item, three-factor model provided the best indices of goodness of fit for CAMI (RMSEA = 0.04, CFI = 0.90, TLI = 0.89). MAKS converged with both CAMI and RIBS. Internal consistency was good for the RIBS and acceptable for CAMI and MAKS. Test–retest reliabilities were excellent for RIBS and poor for CAMI and MAKS, but kappa scores for inter-rater agreement were relatively low for these scales. Results support validity of the original MAKS and RIBS scale and a modified CAMI scale and suggest that stigma is not an enduring trait in this population. The low kappa scores are consistent with first kappa paradox which is due to adjustment for agreements by chance in case of marginal prevalence values. Conclusions Kiswahili versions of the MAKS, RIBS and a modified version of the CAMI are valid for use in the study population. Stigma against people with mental illness may not be an enduring trait in this population.