Availability and rational use of drugs was assessed in a random sample of 19 peripheral health units in two governorates in Democratic Yemen in which an essential drugs programme has been operational for the past few years. Findings were compared with those from seven health units in one governorate in which no such programme had been started. On average, 27 essential drugs were available in the programme area, compared with 17 in the control area. Programme areas carried on average 1 non-essential drug, compared with 17 in control areas. Average stock was adequate for 4 weeks in programme areas and for 1 week in control areas. Health workers in the programme area scored slightly, but not significantly, better in a test on theoretical knowledge on rational drug use. However, programme areas differed considerably from control areas in patterns of drug use, with fewer injections (24.8% vs 57.8% of prescriptions) and fewer antibiotics (46.3% vs 66.8%) being prescribed in programme areas, which also had fewer drugs per prescription (1.5 vs 2.4). The programme has significantly improved the availability and rational use of essential drugs in peripheral health units.