Forty new patients with elevated platelet counts (greater than 600 x 10(9)/l) and without palpable splenomegaly were assigned to diagnostic groups defined by essentially conventional criteria after 3 months follow up: Proliferative (17), reactive (17) or unclassified (six). Mean platelet volume (MPV), platelet distribution width (PDW), platelet nucleotide ratio (ATP:ADP), unstimulated BFU-E derived colony formation from peripheral blood, spleen scan and clinical ischaemia were assessed at the outset, with a view to defining diagnostic criteria for distinguishing primary thrombocythaemia from reactive thrombocytosis. All except the first variable were significantly associated with diagnostic group (P less than 0.05). A simple scoring system was devised: enlarged spleen on scan, or presence of BFU-E, each scored 2; elevated PDW (greater than 2 SD from mean), elevated ATP:ADP (greater than 4 SD from mean) or presence of clinical ischaemia, each scored 1. Score totals greater than or equal to 3 predicted primary thrombocythaemia, and totals less than 3 suggested reactive thrombocytosis (predictive value 89%). The system correctly predicted diagnosis in four out of four (probably six out of six) patients whose diagnosis was not apparent initially, and thus whose results were not used in constructing the scoring system. Exclusion of BFU-E from the system resulted in only one incorrect prediction in this group.