Background: People with type 1 diabetes who use continuous subcutaneous insulin infusion (CSII, or insulin pump therapy) often remove their pump before extended periods of exercise, but this approach might result in reduced glycaemic control and increased risk of hyperglycaemia and ketogenesis. We aimed to assess the efficacy and safety of a hybrid approach, in which basal insulin delivery was divided between CSII and a daily injection of insulin degludec., Methods: In this single-centre, open-label, proof-of-concept, randomised crossover trial done at the LMC Diabetes & Endocrinology research centre, we recruited physically active and aerobically fit participants aged 18 years or older with type 1 diabetes who were using CSII. Participants were randomly assigned (1:1) by use of a computer-generated sequence to one of two sequences of either usual CSII, involving the continuation of the participant's usual CSII regimen, followed by crossover to hybrid CSII, in which the delivery of the participant's usual daily basal insulin dose was split (50% delivered by CSII and 50% delivered by a once-daily morning injection of 100 U/mL insulin degludec), or the opposite sequence (ie, hybrid CSII followed by crossover to usual CSII). Treatment was not masked to the investigators or participants. For each intervention, participants completed a moderate-intensity and a high-intensity in-clinic exercise session in the first week, followed by four high-intensity and two moderate-intensity home-based exercise sessions in the subsequent 3 weeks. Insulin pumps were suspended or disconnected 60 min before exercise and reconnected immediately after exercise during both treatment regimens. The coprimary outcomes were: (1) time spent in the target control range of 4·0-10·0 mmol/L blood glucose after high-intensity exercise, and (2) time spent in target control range of 4·0-10·0 mmol/L blood glucose after moderate-intensity exercise, measured by continuous glucose monitoring in the 6-h period from the start of the high-intensity and moderate-intensity in-clinic exercise sessions. Outcomes were assessed in a modified intention-to-treat population that included all participants who started both intervention phases and completed all of the in-clinic exercise visits. This trial is registered with ClinicalTrials.gov, NCT03838783, and is complete., Findings: Between May 15, 2018, and March 5, 2019, we assessed 43 patients for eligibility, of whom 31 were randomly assigned to receive the usual CSII regimen (n=14) or hybrid CSII regimen (n=17) in the first phase (before crossover). The analysis population consisted of 24 participants who completed both study phases. Compared with the usual CSII regimen, participants on the hybrid CSII regimen had a significantly longer time in blood glucose range of 4-10 mmol/L during the 6-h period from the start of both moderate-intensity (mean difference 86 min [95% CI 61-147], p=0·005; percentage time in range 64% [SD 35] vs 40% [35]) and high-intensity in-clinic exercise session (60 min [11-109], p=0·01; 66% [32] vs 50% [27]). Participants on the hybrid CSII regimen also showed a higher time in blood glucose range of 4-10 mmol/L during home-based exercise sessions (mean difference 23 min [95% CI -1 to 46], p=0·055), with significantly lower time spent in hyperglycaemia than participants on the usual CSII regimen (mean difference 25 min [2-48], p=0·04). These exploratory outcomes also showed no significant difference in the amount of time spent in hypoglycaemia, nor the number of hypoglycaemic events, between the two interventions. There were three study-related adverse events reported with the usual CSII regimen (two hypotension events and one nausea event) and four with the hybrid CSII regimen (two hypotension events and two nausea events)., Interpretation: A hybrid regimen of injected insulin degludec and CSII (with pump removal during exercise) appears to be safe and effective in adults with type 1 diabetes who exercise regularly. This approach could offer improved glycaemic control immediately after exercise and should be further assessed in a larger-scale randomised trial., Funding: Novo Nordisk., (Copyright © 2020 Elsevier Ltd. All rights reserved.)