Abstract The Syrian conflict has caused enormous displacement of a population with a high non-communicable disease (NCD) burden into surrounding countries, overwhelming health systems’ NCD care capacity. Médecins sans Frontières (MSF) developed a primary-level NCD programme, serving Syrian refugees and the host population in Irbid, Jordan, to assist the response. Cost data, which are currently lacking, may support programme adaptation and system scale up of such NCD services. This descriptive costing study from the provider perspective explored financial costs of the MSF NCD programme. We estimated annual total, per patient and per consultation costs for 2015–17 using a combined ingredients-based and step-down allocation approach. Data were collected via programme budgets, facility records, direct observation and informal interviews. Scenario analyses explored the impact of varying procurement processes, consultation frequency and task sharing. Total annual programme cost ranged from 4 to 6 million International Dollars (INT$), increasing annually from INT$4 206 481 (2015) to INT$6 739 438 (2017), with costs driven mainly by human resources and drugs. Per patient per year cost increased 23% from INT$1424 (2015) to 1751 (2016), and by 9% to 1904 (2017), while cost per consultation increased from INT$209 to 253 (2015–17). Annual cost increases reflected growing patient load and increasing service complexity throughout 2015–17. A scenario importing all medications cut total costs by 31%, while negotiating importation of high-cost items offered 13% savings. Leveraging pooled procurement for local purchasing could save 20%. Staff costs were more sensitive to reducing clinical review frequency than to task sharing review to nurses. Over 1000 extra patients could be enrolled without additional staffing cost if care delivery was restructured. Total costs significantly exceeded costs reported for NCD care in low-income humanitarian contexts. Efficiencies gained by revising procurement and/or restructuring consultation models could confer cost savings or facilitate cohort expansion. Cost effectiveness studies of adapted models are recommended.