Abstract The validity of epidemiological surveys and the success of helminth control programmes based on the diagnosis of infection depend upon an accurate screening procedure. The success of all school-based control programmes, whether they involve prior diagnosis or not, depends on the level of school attendance and school enrolment. The degree to which compliance and school absenteeism may affect estimates of helminth infection and the coverage of treatment was investigated using empirical data from a survey conducted in 3 rural schools in Jamaica. Two sequential stool samples were requested from each child aged 9 to 12 years and screened for the presence and intensity of helminth infection using the Kato thick smear technique. Of the 696 children recruited into the study, 94% agreed to participate but only 90% of these children actually provided a stool sample for diagnosis and only 74% returned the second stool. Children infected with moderate to heavy loads of Trichuris trichiura were less likely to comply fully with the protocol than the uninfected children. They also took longer to comply and were absent from school more often than their uninfected counterparts. Increasing the sampling effort increased the compliance of all infected, and heavily infected, children. By giving them the option to comply, heavily infected children were under-represented by the sampling procedure. This has important implications for the design of control programmes and epidemiological surveys.