Temporal coherence was assessed for 11 limnological variables--water temperature, oxygen, conductivity, alkalinity, pH, colour, calcium (Ca), iron, aluminium, total phosphorus and total nitrogen--between 28 boreal lakes in southern Finland for the winter ice-covered period. The lakes were mainly small (<0.2 km2) and brown-coloured, and located within a circle of 10-km radius. A mean Pearson correlation coefficient for all lake pairs and variables averaged 0.37. Temporal coherence of variables across lake pairs was highest for conductivity, Ca, water temperature and alkalinity. The lake pairs with a direct surface water channel connection had a higher coherence than the lake pairs not connected by a stream. The size of the lake or catchment area had little effect on the coherence between the lakes. Temporal coherence was not strongly related to the difference in water colour (dystrophy) between the lakes. However, between polyhumic lakes (colour > 100 g Pt m(-3)) the coherence was generally higher than between less coloured lakes. Year-to-year variation in limnological characteristics could be partly explained by the variation in local weather. In March, water temperature and chemistry were infrequently related to winter weather, rather they correlated with the weather conditions of the previous autumn, while the ambient late winter weather seemed to have a stronger influence on lake conditions in April. Temporal variation in some variables was related to the atmospheric pressure changes over the North Atlantic (the North Atlantic Oscillation, NAO). Our results suggest that the potential effects of climatic change on lakes can be generalised regionally for brown-coloured dystrophic lakes.