Within the self-determination theory of motivation the concept “motivational soup” refers to motivational regulations forming profiles of accumulated drives towards behavior. Few studies have examined the probability of intervention effects to change such motivational profiles over time, knowledge that might inform future program design to promote sustainable exercise motivation. Participants (N=318) were 279 women and 40 men, aged 23-67 years (Mage=46.7; SD=9.4) consisting of adult members of a web-based step contest provided by their employers. Of the 166 individuals randomly assigned to the experimental group, 85 logged in to the digital intervention platform at least once and were considered treated as intended. This group had access to a web-based digital exercise motivation intervention based on SDT for three weeks. The trial had three measure points; T1 baseline, T2 (3 weeks) and a follow up T3 (6 weeks). To investigate the potential effect of the intervention on the odds of participants to change motivational profiles between T1 and T3 we used Latent Transition Analysis. The intervention had positive main effects on exercise level. A four profile solution showed good quality of classification into the separate profiles (entropy = .92). The profiles were labeled high-high (n=262), high-low (n=26), low-high (n=8), and low-low (n=12), where the label high reflected a more autonomous regulation pattern (missing n=10). Participants in the autonomous profile at T1 had high probabilities of remaining there at T3, but slightly lower probability in the control (82%) than the intervention (95%) condition. Participants in the control condition also had a statistically significant increased probability to belong to the profile with a decreased level autonomy T3 (high-low profile) (OR=4.0, p=.008).These results indicate that this digital exercise motivation intervention can increase the likelihood for participants to sustain autonomous motivation profiles over time.