Qualitative and semiquantitative assessments of oestrogen receptor and progesterone receptor positivity determined on previously immunocytochemically stained slides were performed by eight independent assessors. Concordance between assessments of steroid receptor status was good (24/25, 96%). Interassessor variations in estimates of positive immunostaining levels were high, varying by between 10 and 75% for individual slides. In 2 cases estimates for the same section ranged between 15% nuclei positive and 90% nuclei positive. Wide variations were also recorded for slides stained for progesterone receptors. Results using an assessment procedure combining staining intensity and percentage positivity estimates were also subject to marked discordance. A computerised image analysis system, also used to assess slides gave results similar to the mean manually determined percentage positivity values. It is suggested that quality control of steroid receptor immunocytochemical quantification be considered and that automated image analysis may represent an accurate and valid means of achieving this.