The causes of incomplete remyelination in progressive multiple sclerosis are unknown, as are the pathological correlates of the different clinical characteristics of patients with primary and secondary progressive disease. We analysed brains and spinal cords from 51 patients with progressive multiple sclerosis by planimetry. Thirteen patients with primary progressive disease were compared with 34 with secondary progressive disease. In patients with secondary progressive multiple sclerosis, we found larger brain plaques, more demyelination in total and higher brain loads of active demyelination compared with patients with primary progressive disease. In addition, the brain density of plaques with high-grade inflammation and active demyelination was highest in secondary progressive multiple sclerosis and remained ∼18% higher than in primary progressive multiple sclerosis after adjustments for other plaque types and plaque number (P < 0.05). Conversely, the proportion of remyelinated shadow plaques (P < 0.05) and the overall remyelination capacity (P < 0.01) per brain were higher in primary, compared with secondary, progressive multiple sclerosis. By contrast, there were no group differences in the brain load or frequency of low-grade inflammatory plaques with slowly expanding demyelination. Spinal cord lesion loads and remyelination capacity were also comparable in the two patient groups. Remyelinated areas were more vulnerable than the normal-appearing white matter to new demyelination, including active demyelination in secondary progressive multiple sclerosis. ‘Recurrent’ slowly expanding demyelination, affecting remyelinated areas, and the load of slowly expanding demyelination correlated with incomplete remyelination in both groups. In turn, incomplete remyelination in the spinal cord correlated with higher disease-related disability (determined retrospectively; r = −0.53; P < 0.05 for remyelination capacity versus disease severity). By contrast, such a correlation was not observed in the brain. We propose that regulatory and reparative properties could protect the white matter of the brain in patients with primary progressive multiple sclerosis. These patients may, thereby, be spared symptoms until the spinal cord is affected. By contrast, recurrent active demyelination of repaired myelin could explain why similar symptoms often develop in consecutive relapses in relapsing-remitting/secondary progressive multiple sclerosis. Our data also indicate that slowly expanding demyelination may irreparably destroy normal and repaired myelin, supporting the concept of slowly expanding demyelination as an important pathological correlate of clinical progression.