1. We measured respiration of the larvae of aquatic insects from streams in the Ecuadorian Andes in relation to oxygen saturation at 5, 8, 11, 14 and 17 °C. Polycentropus (Polycentropodidae), Lachlania (Oligoneuriidae), Anchytarsus (Ptilodactylidae) and Anacroneuria (Perlidae) represented genera absent from the highest altitudes, reaching 2720, 2930, 3120, 3450 m a.s.l., respectively, while Claudioperla (Gripopterygidae) and Anomalocosmoecus (Limnephilidae) occurred only above 2900 m a.s.l. Our purpose was to determine whether natural altitudinal limits were reflected in physiological critical points on respiration versus oxygen curves and by the effect of temperature on the ability to oxy-regulate.

2. For all six genera, respiration was affected by oxygen saturation and temperature. Respiration (mg O₂ g^-1 AFDM h^-1) at 70% oxygen saturation (Michaelis-Menten fitted) varied from 2.6 to 7.6 between genera at 17 °C, and from 1.3 to 2.5 at 5 °C. Q₁₀ values for this temperature interval ranged 1.5-2.9 (mean 2.3). The two "high-altitude" genera had higher respiration rates at low temperature and oxygen saturation, and their respiration rate saturated at lower temperatures, than three of the four "low-altitude" genera.

3. The oxy-regulatory capacity (critical points and initial decrease in respiration versus oxygen regressions) varied among genera and was affected by temperature. Lachlania, Claudioperla and Anomalocosmoecus had a higher ability to oxy-regulate at low than at high temperatures, Anacroneuria was not clearly affected by temperature, while Polycentropus and Anchytarsus had a greater oxy-regulatory capacity at high than at low temperature. These results indicate that the ability to oxy-regulate is related to the temperature (altitude) at which species naturally occur.

4. Upper altitudinal limits of the six genera were not reflected in their respiratory performance, because all genera had critical minima of temperature and oxygen saturation much lower than those occurring at the limits of their natural distribution. So, the altitudinal limit could not be attributed to absolute short-term physiological tolerance of low temperature and oxygen concentration.

5. Multiple regressions (based on respiration experiments and previously obtained relationships between water temperature, oxygen saturation and altitude) were used to predict how respiration rates should vary with altitude. At the upper limit of the four "low-altitude" genera, respiration rates were 50-68% of those predicted at the centre of the range. With an arbitrary increase of 400 m above the actual limit, the effect of temperature would be a 13% decrease, and that of oxygen a 2% decrease, in respiration rate of Polycentropus, Lachlania and Anacroneuria, while respiration in Anchytarsus would be reduced by 5% by both factors.

6. It seems that, while the immediate decrease in respiration with increased altitude is caused mainly by a decrease in temperature, the long-term survival of a species at given altitudes might be more affected by oxygen saturation. Further quantitative and long-term studies on survival and recruitment in populations and communities are needed to determine the importance of temperature and oxygen for altitudinal limits of aquatic insects.