We examine monitoring data for 28 years of change in a sub-tropical salt marsh where the hydrology had been minimally modified for mosquito control. This extends the work previously published in 2002 for 14 years of data, analysed by Mike Dale (Dale and Dale 2002 Community Ecol. 3: 19–29). The Minimum Message Length method was used in an unsupervised classification to determine the optimum classes, based on the characteristics of the two dominant plant species: Sporobolus virginicus and Sarcocornia quinqueflora. A question at that time was whether the observed changes were only those of state (or condition) or if they were associated with a change in the underlying saltmarsh processes (dynamics). In the 28-year analysis we have been able to address this issue. The classes were generally similar to those in the 2002 analysis. However, class extinctions occurred over the 28 years and only four classes remained: three were stands of Sporobolus and the other was bare mud. The latter, with mangrove pneumatophores, represented the encroachment of Avicennia mangroves into salt marsh. We suggest that the class extinctions and the final loss of most of the plants represent a change in the processes operating in the marsh. The observed changes may be related to sea level and/or climate changes but future research would be needed to assess this.