In-situ cosmogenic 36 Cl production rates from spallation of Ca and K determined in several previously published calibration studies differ by up to 50%. In this study we compare whole rock 36 Cl exposure ages with 36 Cl exposure ages evaluated in Ca-rich plagioclase in the same 10±3ka lava sample taken from Mt. Etna (Sicily, 38°N). The exposure age of the sample was determined by K–Ar and corroborated by cosmogenic 3 He measurements on cogenetic pyroxene phenocrysts. Sequential dissolution experiments showed that high Cl concentrations in plagioclase grains could be reduced from 450ppm to less than 3ppm after 16% dissolution. 36 Cl exposure ages calculated from the successive dissolution steps of this leached plagioclase sample are in good agreement with K–Ar and 3 He age. Stepwise dissolution of whole rock grains, on the other hand, is not as effective in reducing high Cl concentrations as it is for the plagioclase. 330ppm Cl still remains after 85% dissolution. The 36 Cl exposure ages derived are systematically about 30% higher than the ages calculated from the plagioclase. We could exclude contamination by atmospheric 36 Cl as an explanation for this overestimate. Magmatic 36 Cl was estimated by measuring a totally shielded sample, but was found to account for only an insignificant amount of 36 Cl in the case of the 10ka whole rock sample. We suspect that the overestimate of the whole rock exposure age is due to the difficulty in accurately assessing all the factors which control production of 36 Cl by low-energy neutron capture on 35 Cl, particularly variable water content and variable snow cover. We conclude that some of the published 36 Cl spallation production rates might be overestimated due to high Cl concentrations in the calibration samples. The use of rigorously pretreated mineral separates reduces Cl concentrations, allowing better estimates of the spallation production rates.In the Appendix of this paper we document in detail the equations used. These equations are also incorporated into a 36 Cl calculation spreadsheet made available in the supplementary data.