Background: Adequate malaria diagnosis and treatment remain major difficulties in rural sub-Saharan Africa. These issues deserve renewed attention in the light of first-line treatment with expensive artemisinin-combination therapy (ACT) and changing patterns of transmission intensity. This study describes diagnostic and treatment practices in Mto wa Mbu, an area that used to be hyperendemic for malaria, but where no recent assessments of transmission intensity have been conducted. Methods: Retrospective and prospective data were collected from the two major village health clinics. The diagnosis in prospectively collected data was confirmed by microscopy. The level of transmission intensity was determined by entomological assessment and by estimating seroconversion rates using anti-malarial antibody responses. Results: Malaria transmission intensity by serological assessment was equivalent to < 1 infectious bites per person per year. Despite low transmission intensity, > 40% of outpatients attending the clinics in 2006–2007 were diagnosed with malaria. Prospective data demonstrated a very high overdiagnosis of malaria. Microscopy was unreliable with < 1% of slides regarded as malaria parasite-positive by clinic microscopists being confirmed by trained research microscopists. In addition, many 'slide negatives' received anti-malarial treatment. As a result, 99.6% (248/249) of the individuals who were treated with ACT were in fact free of malaria parasites. Conclusion: Transmission intensity has dropped considerably in the area of Mto wa Mbu. Despite this, most fevers are still regarded and treated as malaria, thereby ignoring true causes of febrile illness and over-prescribing ACT. The discrepancy between the perceived and actual level of transmission intensity may be present in many areas in sub-Saharan Africa and calls for greater efforts in defining levels of transmission on a local scale to help rational drug-prescribing behaviour.