Background: While employment rates have increased in people with multiple sclerosis (PwMS), little is known about the longitudinal trends of work productivity. Objective: To describe the longitudinal patterns of work productivity and examine the factors associated with annual change of work productivity of PwMS. Methods: Study participants were employed participants of the Australian MS Longitudinal Study (AMSLS) followed from 2015 to 2019 with at least two repeated measures (n = 2121). We used linear mixed models to examine if the within-individual variations in MS symptoms are associated with changes in work productivity. Results: The mean annual change in work productivity between 2015 and 2019 was −0.23% (SD = 18.68%). Not the actual severity of symptoms but rather the changes in severity of symptoms that are associated with change in work productivity in the same year. In a multivariable model, every unit increase in mean annual change in ‘pain and sensory symptoms’, ‘feelings of anxiety and depression’, and ‘fatigue and cognitive symptoms’ were independently associated with 2.43%, 1.55% and 1.01% annual reductions in work productivity, respectively. Conclusion: Individual changes in work productivity are largely driven by the changes in symptom severity rather than the absolute severity. Stabilising/improving MS symptoms might improve work productivity.