The effect of the width and shape of the New Zealand filter instrument's passbands on measured total ozone accuracy is determined using a numerical model of the spectral measurement process. The model enables the calculation of corrections for the “bandwidth-effect” error and shows that highly attenuating passband skirts and well-suppressed leakage bands are at least as important as narrow half-bandwidths. Over typical ranges of airmass and total ozone, the range in the bandwidth-effect correction is about 2% in XAD for the filter instrument, compared to about 1% for the Dobson instrument. Abstract The effect of the width and shape of the New Zealand filter instrument's passbands on measured total ozone accuracy is determined using a numerical model of the spectral measurement process. The model enables the calculation of corrections for the “bandwidth-effect” error and shows that highly attenuating passband skirts and well-suppressed leakage bands are at least as important as narrow half-bandwidths. Over typical ranges of airmass and total ozone, the range in the bandwidth-effect correction is about 2% in XAD for the filter instrument, compared to about 1% for the Dobson instrument.