The conventional way to treat a mosquito net with pyrethroid insecticide is to apply a standard dosage every 6-12 months, and to avoid washing the net until just before retreatment. In some places, nets are normally washed much more often than this, and it may then be more appropriate to apply smaller amounts of insecticide after each wash. The choice of strategy must take into account not only biological effectiveness, but also users' perceptions of this effectiveness and their net-washing habits. We used focus groups to compare users' responses to nets treated with different dosages and chemicals. One hundred current net users in urban Dares-Salaam were each given a net that had been pretreated either with permethrin (200 or 500 mg/m2), or with lambdacyhalothrin (3 or 15 mg/m2), or with water. Neither participants nor investigators knew which group had received which treatment. Focus group discussions were held after 2, 8 and 12 weeks. Participants greatly preferred treated nets. Low doses were perceived to be less effective, especially after 8 and 12 weeks. After 12 weeks most participants had washed their nets, despite requests to the contrary Dirty nets were regarded as unhealthy and socially unacceptable. Few participants experienced side-effects or expressed fears about the safety of treatment. We conclude that asking people to refrain from washing their nets is unrealistic. A 'low-dose frequent-treatment' strategy of insecticide application may be more appropriate in the long run. At first, however, low doses give perceptibly inferior protection. An initial high (loading) dose, followed by frequent lower (maintenance) dosages, might solve this problem.