Polypharmacy has been linked to higher risks of hospitalisation and death in community samples. It is commonly present in people with dementia but these risks have rarely been studied in this population. We aimed to investigate associations between polypharmacy and emergency department attendance, any and unplanned hospitalisation, and mortality in patients with dementia. Using a large mental health care database in South London, linked to hospitalisation and mortality data, we assembled a retrospective cohort of patients diagnosed with dementia. We ascertained number of medications prescribed at the time of dementia diagnosis and conducted multivariate Cox regression analyses. Of 4668 patients with dementia identified, 1128 (24.2%) were prescribed 4–6 medications and 739 (15.8%) ≥7 medications. Compared to those using 0–3 medications, patients with dementia using 4–6 or ≥7 agents had an increased risk of emergency department attendance (hazard ratio 1.20/1.35), hospitalisation (hazard ratio 1.12/1.32), unplanned hospital admission (hazard ratio 1.12/1.25), and death within two years (hazard ratio 1.29/1.39) after controlling for potential confounders. We found evidence of a dose response relationship with each additional drug at baseline increasing the risk of emergency department attendance and mortality by 5% and hospitalisation by 3%. In conclusion, polypharmacy at dementia diagnosis is associated with a higher risk of adverse health outcomes. Future research is required to elucidate which specific agents underlie this relationship and if reduction of inappropriate prescribing is effective in preventing these outcomes in dementia.