Real estate rental adjustment models have taken on numerous forms and specifications, but have typically been estimated in both a linear and univariate fashion. However, it is clear that real estate actors, both developers and occupiers, can behave differently at various points of the business cycle, in ways that linear-models may not be able to adequately account for. This paper extends previous work on market analysis and forecasting by using regime switching modelling techniques which have been popularised in contemporary empirical macroeconomic research. Evidence of non-linearity in the rental adjustment process is found in the City of London office market and then explicitly modelled using the smooth-transition regression technique. The nonlinear model describes the in-sample movements of rents better than the equivalent linear model, particularly in the late 1980s and early 1990s.