Confirmed cases of American cutaneous leishmaniasis (ACL) and other dermatological diseases were evaluated in Colombia with a clinical prediction rule independently by 3 types of evaluators: community health volunteer (CHV), practical nurse (PN) and programme physician (PP). The adapted prediction rule included 6 variables based upon clinical-historical information. The screening instrument was a rotating tower of coloured squares, one colour for each variable. A score ranging from 0 to 7, and a cutoff point of > or = 4 was selected for ACL classification (sensitivity 94.3%, specificity 53.3% and efficiency 80.3%). Disease classification, total score, and variable-specific score obtained by CHVs and PNs were compared to those obtained by a PP. The impact on case detection in the study area was assessed. Both types of primary health worker had a high agreement with the PP (sensitivity) on the classification of patients with ACL by score, CHV (92.3%) and PN (93.3%). Case detection of ACL increased 3-fold over that observed one year earlier. This screening instrument and prediction rule, when incorporated into a community surveillance programme for ACL, can facilitate greater case detection and appropriate referral for more-specific diagnostic procedures.