An integrated control programme for apple pests was operated on a 3 ha area of the Appleby Research Orchard, Nelson, from 1962 to 1973, in which control of San Jose scale (Diaspidiotus perniciosus) up to 1972 relied only on natural enemies and disease. Over 300 trees were individually assessed for San Jose scale infestation each winter of 1970 to 1972, using a 0 to 8 rating system. Almost one third of the trees had 0 rating in 1970 but all trees were infested by 1972, when almost half were rated 6-8, the most severe infestation grades. The San Jose scale population in the orchard was sampled fortnightly from November 1970 to December 1971 by recording different stages on 1.65 cm(2) pieces of bark. All stages were aggregated. The densities of first and second instar stages were estimated with precision of <= 0.1 (D = SEM/mean) on most occasions, and an achievable precision for adult females was determined to be D = 0.12. Precision of crawler density estimates was poor and this method of sampling was concluded as impractical. The sampling programme identified three generations of San Jose scale in 1970-1971, with population density increase in each generation, with the exception of the third generation on trees suffering die-back (rated 7 or 8). Biological control by natural enemies and diseases from 1962 to 1972 failed to prevent San Jose scale causing extreme infestation and die-back of the apple trees. This research has indicated the sampling required to provide reliable density estimates of the settled first instar, second instar and adult female stages of San Jose scale. These methods could be combined with pheromone trapping of adult males and sticky trapping of crawlers in population dynamics studies.