The conventional dosages of pyrethroid insecticides on mosquito nets assume that nets will be retreated at6–12 month intervals. However, dosage should be related to washing of nets; if nets are only washed once ortwice a year, their dosage requirements will be different to those which are washed fortnightly. A ‘low-dose, fre-quent-wash’ retreatment system might be technically more appropriate and more affordable where nets arewashed frequently, as they are in Dar es Salaam. Moreover, for use as a domestic insecticide, water-based for-mulations of pyrethroid are preferable to the more commonly used emulsifiable concentrates (ECs). This paperreports laboratory evaluations of three formulations (ECs, Flowable, CS) of three pyrethroids (deltamethrin,lambdacyhalothrin, permethrin). Insecticidal activity was tested using serial bioassays at a range of dosagesusing Anopheles gambiae. The water-based formulations were no less effective than the ECs, even at the lowestdosages. Nets treated with 3 mg/m2and then repeatedly washed and retreated after each wash with either3 mg/m2or 1 mg/m2were subjected to gas chromatography analysis. This showed that the amounts ofpyrethroid in the nets accumulated rapidly over the first few wash-retreatment cycles and then remained fairlystable over subsequent cycles. These nets gave consistently high bioassay mortalities throughout the experi-ment, while the mortality declined rapidly after several washes with the nets that were treated at 3 mg/m2butnot retreated. Experimental huts were used to compare the effectiveness of these 2 net retreatment regimes andnets which were not retreated. All nets caused high mortality rates amongst Anopheles females, but had negligi-ble effects on culicines; either in killing them or in preventing feeding. Therefore use of a high ‘loading’ dose forinitial treatment with lower ‘maintenance’ doses for retreatment may be preferable to ensure that net userspromptly perceive the benefits of the insecticide against culicines.