Total ozone measurements using a Dobson spectrophotometer have been performed on a regular basis at Reykjav¿k (64¿08'N, 21¿54'W), Iceland, since 1957. The data set for the entire period of observations has been critically examined. Due to problems related to the calibration of the instrument the data record of ozone observations is divided into two periods in the following analysis (1957--1977 and 1977--1990). A statistical model was developed to fit the data and estimate long-term changes in total ozone. The model includes seasonal variations, solar cycle influences, quasi-biennial oscillation (QBO) effects, and linear trends. Some variants of the model are applied to investigate to what extent the estimated trends depend on the form of the model. Trend analysis of the revised data reveals a statistically significant linear decrease of 0.11¿0.07% per year in the annual total ozone amount during the earlier period and 0.30¿0.11% during the latter. The annual total ozone decline since 1977 is caused by a 0.47¿0.14% decrease per year during the summer with no significant change during the winter or fall. On an annual basis, ozone varies by 3.5¿0.8% over a solar cycle and by 2.1¿0.6% over a QBO for the whole observation period. The effect of the 11-year solar cycle is particularly strong in the data during the early months of the year and in the westerly phase of the QBO. The data also suggest a strong response of total ozone to major solar proton events. Comparisons of the total ozone data made at Reykjav¿k during the last 13 years with total ozone as measured by the total ozone mapping spectrometer (TOMS) on the Nimbus 7 satellite show an excellent agreement for solar zenith angles less than 80¿ at times of satellite overpass. Comparisons with results of the zonally averaged TOMS data at this latitude suggest a significant longitudinal dependence of the linear trends in stratospheric ozone. This is in accordance with recent analysis of the satellite data. ¿ American Geophysical Union 1993