Groundwater contaminated with geogenic arsenic (As) is frequently used as drinking water in Burkina Faso, despite adverse health effects. This study focused on testing low-cost filter systems based on zero-valent iron (ZVI), which have not yet been explored in West Africa for As removal. The active ZVI bed was constructed using small-sized iron nails, embedded between sand layers. Household filters were tested for nine months in a remote village relying on tube well water with As concentrations of 400-1350 μg/L. Daily filtered volumes were 40-60 L, with flow rates of ~10 L/h. In parallel, downscaled laboratory filter columns were run to find the best set-up for optimal As removal, with special attention given to the influence of input pH, flow rate and water/nail contact time. Arsenic removal efficiencies in the field were 60-80% in the first six months of operation. The laboratory experiments revealed that trapped air in the nail layer greatly lowered As removal due to preferential flow and decreased water/nail contact time. Measures taken to avoid trapped air led to a partial improvement in the field filters, but effluent As remained >50 μg/L. Similar structural modifications were however very successful in the laboratory columns, where As removal efficiencies were consistently >95% and effluent concentrations frequently 1000 μg/L. A constantly saturated nail bed and careful flow control is necessary for optimal As removal. Slow flow and longer pauses between filtrations are important for sufficient contact times and for transformation of brown amorphous Fe-hydroxides to dense magnetite with incorporated As(V). This preliminary study has shown that nail-based filters have the potential to achieve As removal >90% in a field context if conditions (filter bed saturation, flow rate, pauses between filtrations) are well controlled.