This study was designed to compare the costs of a pharmacy-based Central Intravenous Additive Service (CIVAS) with those of traditional ward-based preparation of intravenous doses for a paediatric population. Labour costs were derived from timings of preparation of individual doses in both the pharmacy and ward by an independent observer. The use of disposables and diluents was recorded and their acquisition costs apportioned to the cost of each dose prepared. Data were collected from 20 CIVAS sessions (501 doses) and 26 ward-based sessions (30 doses). In addition, the costs avoided by the use of part vials in CIVAS was calculated. This was derived from a total of 50 CIVAS sessions. Labour, disposable and diluent costs were significantly lower for CIVAS compared with ward-based preparation (p < 0.001). The ratio of costs per dose [in 1994 pounds sterling] between ward and pharmacy was 2.35:1 (2.51 pounds:1.07 pounds). Sensitivity analysis of the best and worst staff mixes in both locations ranged from 2.3:1 to 4.0:1, always in favour of CIVAS. There were considerable costs avoided in CIVAS from the multiple use of vials; the estimated annual sum derived from the study was 44,000 pounds. In addition, CIVAS was less vulnerable to unanticipated interruptions in work flow than ward-based preparation. CIVAS for children was more economical than traditional ward-based preparation, because of a cost-minimisation effect. Sensitivity analysis showed that these advantages were maintained over a full range of skill mixes. Additionally, significant savings accrued from the multiple use of vials in CIVAS.