This study determines the optimal number of faecal samples that should be examined in order to minimize the occurrence of false-negative results in the diagnosis of cryptosporidiosis using routine techniques. A total of 23,023 faecal samples from 10,870 patients submitted for parasitological examination were processed by the formalin-ethyl acetate concentration technique and stained using a modified Ziehl-Neelsen method. Cryptosporidiosis was diagnosed in 232 patients (2.13%), 44 of whom were infected by HIV, the prevalence rate in this population group being 15.54%. The increase in the number of diagnoses obtained by the examination of n and (n-1) specimens was evaluated statistically. This study found that three is the optimal number of faecal samples that should be examined when cryptosporidiosis is suspected in immunocompetent patients; whereas, only two samples are required for diagnosing this protozoosis in AIDS patients.