The effect of storage at various temperatures on the bacterial community of a cooling-water sample and a tap-water sample was determined. Samples were stored at 4, 10, 20 and 30 degrees C for 24, 48, 72 and 216 h and the aerobic plate count and bacterial community structure of each were determined using R2A and R3A agars. The culturable count (aerobic plate count on R2A/R3A agar) in both samples varied over time, even after 24 h storage at 4 degrees C, showing that bacterial communities in water are dynamic, even at refrigerator temperatures. At 4 degrees C the culturable count of cooling water initially decreased, followed by a tenfold increase. The tap-water count decreased at 4degrees C. At 10degrees C the pattern was similar. At 20 and 30 degrees C there was a tenfold increase in the culturable count of the cooling water, even after 24 h. In the cooling-water sample, the dominant isolates throughout were Pseudomonas stutzeri and an unidentified Gram negative pink isolate. This isolate was not detected in previous studies where Std I nutrient agar was used. Possibly this isolate plays an important role in cooling-water ecology, but does not grow on the conventional agars. The other isolates appeared randomly on the agar plates. The tap-water sample showed great variation in dominance of species over time. No direct tendencies of rate of decrease or increase could be detected in any of the samples, either in the culturable count or in community structure. Therefor results of analysis after storage cannot be adapted by a pre-determined factor. They must be interpreted with extreme caution, as they do not of necessity reflect the bacterial composition of the sample as drawn, both in terms of total numbers and in terms of community structure. Only counts performed on fresh samples yield reliable results on the total culturable count, and only community structures performed immediately, reflect the state of the community in the system from which the sample was drawn.