The retention and release of total coliforms and Escherichia coli was investigated in hand-pumps removed from tubewells tapping a faecally contaminated aquifer in Matlab, Bangladesh, and from a new hand-pump deliberately spiked with E. coli. All hand-pumps were connected to reservoirs of sterile water and flushed. Faecal coliforms were observed in the discharge from all three of the previously used hand-pumps, at concentrations comparable to levels measured in discharge when they were attached to the tubewells. During daily flushing of one of the previously used hand-pumps, the concentration of total coliforms in the discharge remained relatively constant (approximately 10³ MPN/100 mL). Concentrations of E. coli in the pump discharge declined over time, but E. coli was still detectable up to 29 days after the start of flushing. In the deliberately spiked hand-pump, E. coli was observed in the discharge over 125 days (t₅₀ = 8 days) and found to attach preferentially to elastomeric materials within the hand-pump. Attempts to disinfect both the village and new hand-pumps using shock chlorination were shown to be unsuccessful. These results demonstrate that hand-pumps can act as persistent reservoirs for microbial indicator bacteria. This could potentially influence drinking water quality and bias testing of water quality.