In order to investigate the effects of climate on the shell colour and banding polymorphism of Cepaea nemoralis, surveys were carried out, in 1962 and 1963, in two extensive valley complexes in the Pyrenees; populations were sampled over a wide altitudinal range, such that considerable differences in the climatic environment were apparent, both within and between the valley complexes. The valleys selected for study were that of the Garonne in the central Pyrenees and on the north side of the watershed, and that of the Segre and its major tributary, the Valira, nearer the Mediterranean and on the southern side. The unbanded phenotype exhibits a parallel frequency distribution in colonies sampled near the rivers in both regions. It is at higher frequencies at high altitudes, and also in the lower parts of the valleys sampled, intermediate altitudes having populations which are predominantly fivebanded. There is, however, a marked difference between the two regions in the extent of the changes between the intermediate-altitude banding zones and the lower and upper zones; unbanded reaches much higher frequencies regularly in the upper zones of the Garonne compared with those in the Segre-Valira, whereas it is the latter region which exhibits really high frequencies in the lower zone. The distribution of frequencies of the yellow morph in the Garonne resembles that of unbanded; only intermediate-altitude populations have appreciable frequencies of pink, other populations having high or very high frequencies of yellow. In the Segre-Valira, there are some parallels in frequency distribution, but yellow is at a high overall frequency almost everywhere. The species is polymorphic for lip colour in the Pyrenees, and the results agree with Lamotte's for other Pyrenean valleys; the white-lipped morph is rare or infrequent in the foothills, and reaches high frequencies in the higher parts of the mountains. The relative constancy of morph frequencies over considerable distances, and their regular distribution between valleys and between the two regions (for example) show that selection is operating. It is argued that the colour and banding zones are area effects as there is no evidence for variation with habitat in districts where this can be tested by comparing open habitat samples with samples taken from woods, despite the occurrence of visual predation; and certain morph frequencies characterize quite clearly geographical areas rather than habitats. The broad correlation between morph frequencies--particularly banding morph frequencies-and altitude in the upper parts of the valleys suggested that some climatic aspects might be acting selectively. Comparison of samples made from hill slopes with those made by the rivers shows that the former tend to have less extreme morph frequencies; thus, at similar altitudes, hillside populations have lower frequencies of unbanded and of white-lip, and tend to have lower frequencies of yellow. This indicates that the factors favouring these morphs at higher altitudes do not act at the same strength on the hill slopes as in the valleys. It seems likely, from the accepted generalization that hill slopes are usually more temperate than valley bottoms, that cold air and concomitant low temperatures favour unbanded, yellow, and white-lip. Consideration of aspect of the valleys, and of certain colonies exceptional for banding frequencies concur with this: north-facing valleys and occluded situations tend to have populations with higher frequencies of unbanded than south-facing ones and sunny open localities. A comparison of morph frequencies in the Garonne and Segre-Valira regions with their climates also supports this interpretation for colour and banding. Similarly, the increase in frequencies of unbanded and of yellow at lower altitudes in the Garonne, and the very marked rise in frequency of unbanded in the Segre-Valira suggests that these two morphs are also favoured in conditions which tend to be warm and dry. The extent of the changes agrees, for the lower Segre-Valira, unlike the lower surveyed reaches of the Garonne, is essentially mediterranean in climate, and the frequency of unbanded is overall much higher there. Other field evidence, both from the Pyrenees and elsewhere, is in broad agreement with yellow and unbanded being favoured in more extreme climatic conditions, and most of the artificial resistance experiments performed with Cepaea accord in that these two morphs survive better when subjected to severe temperatures. The results are discussed with relevance to morph frequency distributions elsewhere; it is suggested that area effects in unbanded and yellow may occur when climate becomes sufficiently extreme, and some evidence is presented that this may be so on high country in the English counties Of Staffordshire and Cumberland. It is also suggested that if the strength of visual selection is relaxed in open country, as recent work indicates, then area effects in pink and fivebanded may occur in mild conditions; some evidence is produced in support of this idea. The clines in banding in the lower valleys are stepped; banding morph frequencies change apparently suddenly from one relatively stable frequency to another in both regions, without relation tO topographical features which might cause climatic disjunction in the vicinity of the step. It is tentatively suggested that co-adaptive selection may interact with climatic selection in the intermediate and lower zones, to produce sharp steps in frequency with stability despite change in altitude and climate on either side.