We examine inorganic chlorine (Cly) partitioning in the summer lower stratosphere using in situ ER-2 aircraft observations made during the Photochemistry of Ozone Loss in the Arctic Region in Summer (POLARIS) campaign. New steady state and numerical models estimate <ClONO2>/<HCl> using currently accepted photochemistry. These models are tightly constrained by observations with OH (parameterized as a function of solar zenith angle) substituting for modeled HO2 chemistry. We find that inorganic chlorine photochemistry alone overestimates observed <ClONO2>/<HCl> by approximately 55--60% at mid and high latitudes. On the basis of POLARIS studies of the inorganic chlorine budget, <ClO>/<ClONO2>, and an intercomparison with balloon observations, the most direct explanation for the model-measurement discrepancy in Cly partitioning is an error in the reactions, rate constants, and measured species concentrations linking HCl and ClO (simulated / too high) in combination with a possible systematic error in the ER-2 ClONO2 measurement (too low). The high precision of our simulation (¿15% 1&sgr; for <ClONO2>/<HCl>, which is compared with observations) increases confidence in the observations, photolysis calculations, and laboratory rate constants. These results, along with other findings, should lead to improvements in both the accuracy and precision of stratospheric photochemical models. ¿ 2001 American Geophysical Union