This study asks whether instructed response items validly detect careless respondents. We found that instructed response items have high internal consistency, good convergent validity with other types of carelessness detectors (synonyms, antonyms, and odd-even index), and good discriminant validity with acquiescence. Excluding careless respondents based on increasingly strict cutoff values leads to stronger negative correlations between regular and reverse-keyed items. Overall, the results favor the continued use of instructed response items to identify careless respondents.