Objective To evaluate the associations of self-reports of suicidal ideation and behavior using the Columbia-Suicide Severity Rating Scale (C-SSRS) in a survey of patients receiving mental health services in the Veterans Health Administration (VHA) with reports of attempts documented in medical records and administrative data. Method The C-SSRS was administered to 15,373 Veterans in the Veterans Outcome Assessment (VOA) survey. Concurrent validity was evaluated by comparing self-reports from the past 3 months with VHA records. Predictive validity was evaluated by logistic regression models using attempts over the subsequent 3 months as the outcome. Results Tests of concurrent validity found strong associations between self-reports and attempts documented in VHA records, but there were substantial numbers of discordant responses. In tests of predictive validity, area under the ROC curve for predicting future attempts was >0.8. There were differences in the distribution of responses and of psychometric properties across VHA mental health programs. Conclusions Findings support the value of screening and the validity of the self-reports based on the C-SSRS, but limitations in concordance with medical records and variability across programs suggest the need for clinical judgment in interpreting responses.