Summary: Objective To determine the efficiency of routine tuberculosis (TB) case detection by examining sputum smear positivity for acid-fast bacilli in relation to duration of cough, characteristics of TB suspects examined and health service factors.Method We combined patient interviews with routine data from laboratory registers in 6 health care facilities in San Juan de Lurigancho district, Lima, Peru. A TB case was defined as a TB suspect with at least one positive sputum smear. We calculated adjusted odds ratios with 95% confidence intervals for the association between smear positivity and health service and patient's characteristics.Results Smear positivity was 7.3% (321/4376). Of the 4376 adults submitting sputa, 55.3% (2418) reported cough for <14 days. In this group, smear microscopy yielded 3.2% (78/2418) positive results vs. 12.4% (243/1958) in patients coughing for 14 or more days. Having cough for >2 weeks, being referred by health care staff, attending a secondary-level health care facility, male sex and age between 15 and 44 years were independent determinants of smear positivity.Conclusions Routine case detection yields a low proportion of smear-positive cases because of the inclusion of a high proportion of patients without cough or coughing for <2 weeks. Adherence to the national TB control programme guidelines on the selection of TB suspects would have a positive impact on the smear positivity rate, reduce laboratory costs and workload and possibly improve the reading quality of smear microscopy. © 2010 Blackwell Publishing Ltd.