Observations of marine air temperature (MAT) by Voluntary Observing Ships (VOS) are known to contain significant biases due to solar heating of the sensor environment. MAT and humidity observations are usually made using wet- and dry-bulb thermometers housed in Stevenson screens, or with psychrometers. These instruments are typically mounted in the bridge wings or on the wheel-house top. If not sited carefully then the instruments can be poorly exposed to the undisturbed environmental conditions and have inadequate ventilation, leading to biased observations of both MAT and humidity. In this paper we use observations collected as part of the VOS Climate (VOSClim) project to investigate the relationship between instrument exposure and heating errors. The heating errors are estimated as the difference between the observed MAT and the collocated output of a numerical weather prediction model. The instrument exposures are assessed from hotographs of the instruments. Currently, photographs of the instruments and sufficient observations exist for 17 VOSClim ships. Two methods of assessing the instrument exposure using the observations are presented. The first method is based on the skewness of the distribution of estimated heating errors for individual ships. The second method is based on a correction developed to correct the heating errors and uses the ratio of the heating to cooling terms in the correction. When ships are ranked both on the skewness and on the ratio of the heating to cooling terms, there is a statistically significant correspondence between the rankings and the visual assessments of instrument exposure. The skewness of the distribution of estimated errors in MAT is proposed as a simple indicator of instrument exposure.