Track-etch detectors made of cellulose nitrate (LR 115, Kodak Pathé) and polycarbonate (CR 39, Pershore Mouldings Ltd) were compared regarding sensitivity and background when used as detectors for boron determination in biological samples. Measurements were made on two kinds of sample, cryosectioned biological tissue, and liquid samples deposited directly on the detector surface as microdroplets. The CR 39 films were pretreated or washed before irradiation. When cryosectioned tissue was used, measurements were made with and without the inclusion of Mylar foils between the samples and the detectors. Foil thicknesses used were 2 microns in the case of LR 115 and 2, 4, and 6 microns in the case of CR 39. All samples were irradiated with a thermal-neutron fluence of 5 x 10(12) neutrons cm-2 at the thermal-neutron facility in Studsvik, Sweden. The use of a Mylar foil generally suppressed the background tracks relative to the tracks from the 10B disintegration. No difference in resolution between CR 39 and LR 115 was observed. Pretreatment of the CR 39 resulted in an improved sensitivity of detection but the detector became saturated at 0.25 parts per million of 10B. The background was found to be lower in the pretreated detector.