A sound presented temporally proximal to a target sound has effects on intensity resolution and loudness. The aim of this work was to develop a model for three phenomena. First, an intense non-simultaneous masker causes a large increase in intensity difference limens for a midlevel standard, but has only a small effect at low and high standard levels. Second, a non-simultaneous masker introduces a systematic shift in target loudness towards masker loudness. For example, loudness of the target is enhanced by a masker higher in level than the target. Third, in rhythmic equitone sequences constructed as pairs of sounds separated by a longer interval, an accent is perceived on the second sounds of the pairs, i.e., loudness of the second sounds exceeds loudness of the first sounds. Previous studies indicated a close relation between the masker-induced changes in loudness and intensity resolution. The model formulated in this thesis attributes the empirical similarities between the first two phenomena to a common mechanism. Above that, the question was studied whether the interval-produced accent can be understood as loudness of the second tone of a pair being enhanced by the first tone. Based on a review of data and models, a new model for loudness and intensity resolution in non-simultaneous masking was developed, termed the Similarity Model. It assumes that the memory representation of target loudness used in a loudness matching task is a weighted average between target and masker loudness [Elmasian, Galambos, & Bernheim (1980), J. Acoust. Soc. Am. 67, 601-607]. The important new feature of the model is an effect of the target-distractor similarity. The weight assigned to masker loudness is assumed to be a function of the perceptual similarity between masker and target, decreasing with the perceptual difference between the two sounds. Based on an idea by Carlyon and Beveridge (1993, J. Acoust. Soc. Am. 93, 2886-2895) the model predicts the elevation of the just-noticeable difference (jnd) to be monotonically related to the loudness change caused by a non-simultaneous masker. Unlike previous models, the similarity model predicts a mid-difference hump, i.e., at each target level, the effects of the masker are expected to be most pronounced at intermediate masker-target level differences. The model was tested in three experiments. In Experiment 1, intensity difference limens were measured for 25, 55, and 85-dB SPL standards, while the masker-target level difference was varied across a range of −60 to +60 dB. Compatible with the predictions of the similarity model, listeners produced a mid-difference hump and the masker caused significant jnd elevations at the lowest and the highest standard level. Previous models predict no or only very small effects of the masker at low and high standard levels, and at midlevels a monotonic increase of the jnd with masker level. The data are incompatible with these models. In Experiment 2, loudness matches were obtained in the same conditions as in Experiment 1 and for the same listeners. Mid-difference humps were observed again. The masker-induced loudness change was smallest at the lowest target level. In Experiment 3, similarity on the duration dimension was varied. For masker and target additionally differing in duration, the similarity model predicts the same pattern (i.e., a mid-difference hump), but a generally smaller effect of the masker. Results were compatible with this prediction. According to the similarity model, the common mechanism underlying both loudness enhancement and the jnd elevation is the inclusion of masker intensity information in the decisions made in the respective experiments. Compatible with this assumption, the correlation between loudness enhancement and the difference limen was significant. The correlation was rather weak, however, partly because loudness decrement corresponded to small jnd's. It was demonstrated that the similarity model can quantitatively predict the loudness change caused by a non-simultaneous masker. The mathematical modelling combines properties of the auditory periphery, represented by the loudness function for pure tones, with the presumably more centrally located influence of masker loudness on the memory representation of target loudness. With only two free parameters, the model produced reasonable to excellent fits to previous data sets and to the data from Experiment 2. The smaller amounts of loudness enhancement found at the lowest target level were correctly predicted, due to the steeper slope of the loudness function at low levels. A signal detection approach was used to model the effects the inclusion of masker intensity information has on discrimination performance. The predicted patterns were not compatible with the data, however.