Objective  Measure effectiveness of intermittent slow sand filtration for reducing child diarrhoea among households using unimproved water sources in rural Kenya.

Methods  A randomized controlled trail was conducted among populations meeting a high‐risk profile for child diarrhoea from drinking river water in the River Njoro watershed. Intervention households (30) were provided the concrete BioSand Filter and instructed on filter use and maintenance. Control households (29) continued normal practices. Longitudinal monthly monitoring of diarrhoea (seven‐day daily prevalence recall) and of influent, effluent, and drinking water quality for fecal coliform was conducted for 6 months.

Results  Intervention households had better drinking water quality than control households (fecal coliform geometric mean, 30.0 CFU vs. 89.0 CFU/100 ml, P < 0.001) and reported significantly fewer diarrhoea days (86 days over 626 child‐weeks) compared to controls (203 days over 558 child‐weeks) among children up to 15 (age‐adjusted RR 0.46; 95 % CI = 0.22, 0.96). Greater child diarrhoea reduction due to the intervention (age‐adjusted RR 0.23, 95 % CI = 0.10, 0.51) was observed among the sub‐group using unimproved water sources all of the time.

Conclusion  Intermittent slow sand filtration, a non‐commercial technology, produces similar observed effects on child diarrhoea as commercial POU products, adding to the range of effective options for poor populations (chlorination, ceramic filtration, solar disinfection, flocculation/disinfection).