The modelling results of tropospheric ozone concentration in the urban area of Cali, Colombia, suitable for short term forecasting, are presented. Results were obtained by using an univariate time series analysis. The method was applied to a series of 2496 hourly ozone concentration data from one of the city’s air quality monitoring network stations. Data were collected from April to July of 2003. A total of 104 consecutive days were covered: the first 93 days were used for model estimation, and the remaining 11 days for model validation. This technique can be used to predict (up to 8 hours) in advance high-ozone levels, allowing the environmental authorities to issue alerts to the population for possible air-quality impact on health.