The NACP electron chamber is one of three parallel-plate chambers recommended for use in the UK. Measurements with this chamber type have indicated a problem in determining the recombination correction. This is due to a variation of the ionization current I with polarizing voltage V which deviates from the accepted Boag theory. It is shown that there is a chamber-dependent threshold voltage below which the NACP chamber follows the Boag theory. Above this voltage the chamber should be used with caution, although it is still possible to correct for the dependence of the chamber response on the dose per pulse. The existence of such deviations from theory demonstrates the usefulness of the 1/ I against 1/ V plot and the limitations of the Boag two-voltage analysis. Values for the initial recombination and the coefficient of general recombination are measured for several NACP chambers. It is shown that from these one can derive a value for the effective plate separation and the collector radius of each chamber. Differences in the behaviour of NACP chambers manufactured by Scanditronix and Dosetek are discussed and the implications of free-electron collection are considered.