A population study (n=114) was performed. Fifty blood donors were selected to provide a distribution, slightly biased from normal, favour of the GC 1F-1F and GC 1F-1S pheno-types. A protocol was devised for preparing large bloodstains. The longest GC bands were obtained from the edge of a stain after the blood had been treated with K+/EDTA. Each laboratory received a representative portion of the large bloodstains for GC typing. Five of the eight laboratories correctly grouped all the bloodstains. No errors directly attributable to the system were recorded in over 800 tests, indicating that GC in bloodstain can be typed reliably using the combination of isoelectric focusing in ultrathin narrow pH interval gels followed by immunofixation and silver staining. 4 tables and 9 references. (Publisher abstract)