Objectives Antidepressant are commonly prescribed to persons with cognitive disorders to treat depressive and other neuropsychiatric symptoms despite the inconclusive evidence on their effectiveness on this indication. We studied whether recent hospitalisation was associated with antidepressant initiation in people with Alzheimer's disease (AD). Methods The register‐based Finnish nationwide Medication use and Alzheimer's disease cohort includes community‐dwelling persons diagnosed with AD during 2005–2011 in Finland (n = 70,718). This study was restricted to people who initiated antidepressant use after AD diagnosis and had no active cancer treatment and schizophrenia or bipolar disorder diagnoses. We performed a nested case‐control study with antidepressant initiators as cases. A matched noninitiator (sex, age and AD duration), was identified for each initiator (15,360 matched pairs). Recent hospitalisation was defined as hospital discharge within the past 14 days of initiation. Results Antidepressant initiators were four times more likely (adjusted odds ratio: 4.41, 95% confidence interval: 4.06–4.80) to have been hospitalised within the past 2 weeks before initiation (21.2%, n = 3250) than matched noninitiators (5.4%, n = 831) and the duration of hospital stay was significantly longer among initiators. Dementia was the most common main discharge diagnosis among both initiators (43.8%, n = 1423) and noninitiators (24.8%, n = 206). Conclusion Recent hospitalisation was strongly associated with antidepressant initiation in persons with AD. Further studies are needed to investigate whether this is due to neuropsychiatric symptoms leading to hospital admission, inpatient care triggering or worsening neuropsychiatric symptoms or other indications. Nonpharmacological treatments for neuropsychiatric symptoms should be prioritised and the threshold for prescribing antidepressants should be high.