Background The aim of this study was to explore the predictive value of baseline characteristics in relation to changes in physical activity (PA) and sedentary behaviour among diabetic and pre-diabetic patients participating in a primary care based exercise intervention. We used a descriptive case series among diabetic and pre-diabetic patients (n = 119, 50.8% male, mean age 65.5 (SD = 7.8)). Measurements took place with questionnaires at baseline and two years after the start of the intervention. Predictor variables included demographic factors, Body Mass Index, baseline PA and sitting time, and baseline socio-cognitive profile. Results At follow-up, respondents spent more time being physically active than at baseline. For the total group, the average sitting time remained almost unchanged between the two measurements. Further exploration showed that respondents who had relatively high levels of PA at the start of the intervention, increased their total sitting time, while respondents with relatively low levels of PA at the start decreased their sitting time. The socio-cognitive profile did not predict behaviour change. The intervention appeared to be suitable for people with a low-education level, but the results should be interpreted in view of the limitations of the study such as the non-controlled design, self-reported outcomes and selective drop-out of participants. Conclusions Interventions for this specific target group may need to put more emphasis on the prevention of increased sitting time. The finding that the socio-cognitive profile did not predict behaviour change may underline the proposition that decisions to initiate and maintain PA behaviour change are to a large extend non-linear events. Acknowledging the possible non-linearity of the relationship between socio-cognitive determinants and behaviour change will help our understanding of this complex and dynamic process.