Mapping the pleasantness of an urban environment is an alternative approach, closer to the city dweller's perception, than standardized sound levels cartography. This study reports on modeling pleasantness in urban context using perceptual assessments and sound measurements for specific locations during an urban walk. These assessments have been collected from four groups of approximately ten participants on 19 different assessment locations, along a 2,1 km-long path traveled in both directions. Simultaneously, ⅓ octave band sound levels and audio were recorded. Perceptual and physical models of pleasantness are proposed for specific locations based on multiple linear regressions. A multilevel analysis was performed, and it is shown that a perceptual model that includes perceived loudness joined to the perceived time of presence of traffic, voices and birds explains 90% of the pleasantness variance due to the sound environment variations. Physical models that include the original acoustic indicators that are most correlated with perceptual variables explain 85% of this variance. Thanks to these models, a unique averaged pleasantness value is defined for each assessment location from the perceptual or physical collected assessments. The Pearson's correlation coefficient between the averaged perceived pleasantness and the modeled values from perceptual assessment reaches r(19)=0.98, and r(19)=0.97, with the modeled values from physical measurements. These results make it possible to consider the use of this kind of models in a cartographic context. As the path was traveled in both directions, the presentation-order effect has also been assessed, and it has been found that path direction did not have a significant impact on the pleasantness assessment at specific locations, except when very strong sound environment changes occurred. Finally, the study gives some insights about the retrospective global pleasantness assessment for urban walks. For very short walks between two assessment locations, a recency effect is shown. Nevertheless, this effect doesn't seem to be significant when longer routes are assessed.