Light transmission platelet aggregation tests are important for diagnosing platelet function defects. However, uncertainties exist about the best procedures to determine aggregation reference intervals. We investigated methods for determining reference intervals for light transmission aggregation tests, using the % maximal aggregation values for prospectively collected data on healthy control samples. Reference intervals for samples tested at 250 x 109 platelets/l were determined by mean ± 2 standard deviations and non-parametric analyses. To establish reference intervals for tests on thrombocytopenic subjects, regression analyses were used to estimate 95% confidence limits for % maximal aggregation, according to sample platelet counts, using data for control samples diluted to match the platelet count of undiluted thrombocytopenic patient platelet-rich plasma samples. For samples tested at 250 x 109 platelets/l, non-parametric analyses described 95% of data for healthy control samples better than mean ± 2 standard deviations. For samples tested at lower counts, to match thrombocytopenic samples, the % maximal aggregation was influenced by platelet count and derived limits were wider at very low platelet counts for almost all agonists. With ristocetin, it proved feasible to test samples with very low platelet counts to exclude Bernard-Soulier syndrome and type 2B von Willebrand disease. Non-parametric analyses should be the preferred method to establish light transmission aggregation reference intervals for samples tested at normal platelet counts. The derived limits for thrombocytopenic samples provide guidance for evaluating thrombocytopenic platelet function disorders, including which agonists to test, based on the sample platelet count.