Background and purpose The rate at which the chance of a good outcome of endovascular stroke therapy (EVT) decays with time when eligible patients are selected by baseline diffusion‐weighted magnetic resonance imaging (DWI‐MRI) and whether ischaemic core size affects this rate remain to be investigated. Methods This study analyses a prospective multicentre registry of stroke patients treated with EVT based on pretreatment DWI‐MRI that was categorized into three groups: small Diffusion‐Weighted Imaging Alberta Stroke Program Early Computed Tomography Score (DWI‐ASPECTS) (8–10), moderate (5–7) and large (<5) cores. The main outcome was a good outcome at 90 days (modified Rankin Scale 0–2). The interaction between onset‐to‐groin puncture time (OTP) and DWI‐ASPECTS categories regarding functional outcomes was investigated. Results Ultimately, 985 patients (age 69 ± 11 years; male 55%) were analysed. Potential interaction effects between the DWI‐ASPECTS categories and OTP on a good outcome at 90 days were observed (Pinteraction = 0.06). Every 60‐min delay in OTP was associated with a 16% reduced likelihood of a good outcome at 90 days amongst patients with large cores, although no associations were observed amongst patients with small to moderate cores. Interestingly, the adjusted rates of a good outcome at 90 days steeply declined between 65 and 213 min of OTP and then remained smooth throughout 24 h of OTP (Pnonlinearity = 0.15). Conclusions Our study showed that the probability of a good outcome after EVT nonlinearly decreased, with a steeper decline at earlier OTP than at later OTP. Discrepant effects of OTP on functional outcomes by baseline DWI‐ASPECTS categories were observed. Thus, different strategies for EVT based on time and ischaemic core size are warranted.