If the behavior of long-lived radon progeny was well understood, measurements of these could be used in epidemiological studies to estimate past radon exposure. Field measurements were done in a radon-prone area in the Ardennes (Belgium). The surface activity of several glass sheets was measured using detectors that were fixed on indoor glass surfaces. Simultaneously the indoor radon concentration was measured using diffusion chambers. By using Monte Carlo techniques, it could be proven that there is a discrepancy between this data set and the room model calculations, which are normally used to correlate surface activity and past radon exposure. To solve this, a modification of the model is proposed.