Conventional UV−visible spectroscopy instruments measure the extinction spectrum of solutions in a transmission configuration. Because of the finite (nonzero) acceptance angle in detection, errors due to forward scattering and multiple scattering can be introduced when measuring scattering samples. We here experimentally quantify these errors using polystyrene spheres of different sizes for two representative analytical/research UV− visible instruments, one based on a single-beam diode array and the other on a double-beam scanning configuration. The measured spectra for particles larger than 1 μm are shown to differ between the two instruments, even at low concentrations, and also vary with concentration (in contradiction with the Beer−Lambert law). We show that systematic errors in the range of 10−40% are common in such measurements. We propose a model accounting for both forward-and multiple-scattering errors and demonstrate its agreement with our experimental results. This model could reduce systematic errors in measurements of scattering samples by up to 40%.