In situ, quantitative infrared spectrometry has been used to measure the evolution of carbon dioxide during the photo-degradation of pigmented and unpigmented acrylic-emulsion paint films. It is demonstrated that the method permits a fast and convenient way of assessing the resistance of such paints to UV degradation. However, it is also shown that valid comparisons of paints prepared with different grades of titanium dioxide, but with the same acrylic emulsion, require careful selection and control of the experimental conditions. Thus, the relative importance of (a) direct photochemical oxidation of the binder and (b) indirect photo-catalysed oxidation (mediated by titanium dioxide) depend not only on the spectral distribution of the radiation source, particularly the output between 300 and 400 nm, but also on the humidity of the ambient atmosphere. The increased rate of oxidation with increased humidity is not unexpected, but the size of the effect is unexpectedly high. In addition, there is some carbon dioxide formation during irradiation of the paint film in nitrogen and further investigation of this phenomenon is in hand. For both pigmented and unpigmented films, the rate of carbon dioxide evolution is sensitive to the film thickness. The dependence of the oxidation rate on the thickness of clear, unpigmented films is consistent with the measured UV absorption. However, for pigmented films, this dependence extends over greater film thicknesses than predicted by a simple model based on paint film optics.