Even when relatively infrequent, careless and random responding (C/RR) can have robust effects on individual and group data and thereby distort clinical evaluations and research outcomes. Given such potential adverse impacts and the broad use of self-report measures when appraising addictions and addictive behavior, the detection of C/RR can reduce error substantially. Based on earlier research using a video game questionnaire as an exemplar, we cross-validated promising items for detecting C/RR and developed an expanded set of items, in this case using an Internet questionnaire to examine efficacy and generalization. Research participants were instructed to complete the questionnaire in standard fashion (i.e., cooperatively) or to adopt either a careless or random response set. Careless and random responders often obtained elevated mean questionnaire scores. Most items for detecting careless or random responding demonstrated significant differences across groups, and combinations of items showed high levels of accuracy, particularly in detecting random responders. Results suggest that a relatively small number of items, which might only add a minute or two to questionnaire completion time, can detect the great majority of random responders and most careless responders. Guidelines are provided for the development and application of items.