We have measured 36Cl in a suite of 33 rock samples having well-constrained exposure histories and ages. The 36Cl production parameters were estimated by minimizing the squared deviations between the 36Cl and independent ages, yielding the following production parameters: spallation and muon production from Ca 2940¿200 atoms 36Cl (mole Ca)-1 yr-1, spallation from K 6020¿400 atoms 36Cl (mole K)-1 yr-1, and neutron production in air 586¿40 fast neutrons (g air)-1 yr-1. The new production constants for spallation on Ca and thermal neutron activation are in good agreement with previous results, but that for spallation on K is about 50% larger. The past decade has seen a rapid growth in the application of terrestrial cosmogenic nuclides produced by cosmic-ray particles within rocks at the earth surface. Nuclear cross sections, particularly for reactions initiated by neutrons and muons, are not known with sufficient accuracy so production rates for these nuclides must be empirically calibrated using independently dated geological samples. Direct production estimates for 36Cl were first obtained by Zreda et al. (1991), who measured 36Cl in mineral separates from granitic boulders on moraine crests at Chiatovitch Creek in the eastern White Mountains of California and Nevada. This data set was the best available at the time, but since then several factors have prompted a reevaluation of the 36Cl production rates. One is that Zreda et al. (1994) have shown that there can be significant scatter of the cosmogenic ages of individual glacial boulders toward ages younger than that of the moraine, due to soil erosion. A second factor is that a much wider range of samples is now available for this purpose. Finally, recent theoretical advances in calculating production by thermal neutron absorption (Liu et al., 1994) now permit more realistic calibration of this reaction. ¿ American Geophysical Union 1996