SYNOPSIS Range expansion, achieved by insertion of a variable resistance between the colorimeter and the recorder together with the use of 550 nm colorimeter filters, has resulted in markedly improved sensitivity for antibody detection, and improved sample identification, in a low ionic strength-protamine sulphate (LISPS) system. Range expansion also permits a lower concentration of red cells to be used, thus economizing on fully typed cells. Glycerol stored frozen cells were found to be only slightly less sensitive than fresh cells in this system.An automated antibody screening system should detect all antibodies of clinical significance with a sensitivity at least equal to and preferably better than the manual method it replaces. The bromelinmethyl cellulose (BMC) system described by Marsh et al (1968) has the disadvantage of not detecting weak antibodies to the enzyme degradable antigens MNS and Fya, and in our hands did not detect weak anti-K. Our attention was drawn to the low ionic strength-polybrene/protamine sulphate system described by Lalezari (1967;1968).In common with Dawes and Goldsmith (1975), we found that polybrene caused the aggregated cells to adhere to the glass coils, owing to the high degree of surface absorption of polybrene to glass, as noted by Greenwalt and Steane (1973). This resulted in cell baseline fluctuations, excessive 'carry-over', and poor sample identification. Reduction in the polybrene concentration to overcome these problems resulted in a loss of sensitivity. These difficulties were not experienced when protamine sulphate (Lalezari, 1967;Judd, 1971) was used, and this became the reagent of choice. Experiments revealed that the LISPS system enhanced the reaction of MNS and Fya antibodies. Anti-K gave reactions equivalent to manual antiglobulin tests, but the system was slightly less sensitive than the BMC system in detecting weak anti-D, thus confirming the findings of Perrault and Hogman (1971).