The influence of the condensation instrument on the behaviour of amalgam restorations was investigated in a clinical study for which in a group of forty-nine patients, 125 Class I or II amalgam restorations were made by each of two operators. The restorations were made in series of five per patient. For each restoration within a series, the amalgam was condensed with a different instrument (hand condensation, Amalgam Condensaire, Speedomatic, Bergendal Vibrator and Electro-mallet). For a period of 2.5 years the restorations were checked at half-year intervals for failures. The marginal integrity of the restorations was assessed half-yearly with a six-point photo rating scale. The discolouration, roughness formation and the marginal integrity of the restorations were assessed clinically with four-point rating scales. The number of failures and the assessments were analysed for the effect of the condensation instrument, the patient and the operator. After 2.5 years, 12% of the restorations had failed, which is too few for the tracing of a condensation instrument and a patient effect. The operator, however, had an influence on the failure percentage. The condensation instrument had no influence on the clinical behaviour of the restorations, however the patient had. The operator only had an influence on the marginal integrity. This suggests that clinically the condensation results depend more on the operator rather than the instrument.