Ocean surface partial pressure of carbon dioxide (pCO2) is a key factor controlling air–sea CO2 fluxes. Most surface pCO2 data are collected with relatively large and complex air–water equilibrators coupled to stand‐alone infrared analyzers installed on Ships of OPportunity (SOOP‐CO2). This approach has proven itself through years of successful deployments, but expansion and sustainability of the future measurement network faces challenges in terms of certification, autonomy, and maintenance, which motivates development of new systems. Here, we compare performance of three underway pCO2 measurement systems (General Oceanics, SubCtech, and Pro‐Oceanus), including a recently developed compact flow‐through, sensor‐based system. The systems were intercompared over a period of 34 days during two crossings of the subpolar North Atlantic Ocean. With a mean difference from the General Oceanics system of −5.7 ± 4.0 μatm (Pro‐Oceanus) and −4.7 ± 2.9 μatm (SubCtech) during the 1st crossing, our results indicate potential for good agreement between the systems. The study highlighted the challenge of assuring accuracy over long periods of time, particularly seen in a worse agreement during the 2nd crossing, and revealed a number of sources of systematic errors. These can influence accuracy of the measurements, agreement between systems and include slow response of membrane‐based systems to pCO2 changes, “within‐ship” respiration due to biofouling, and bias in measurement of the temperature of equilibration. These error sources can be controlled or corrected for, however, if unidentified, their magnitude can be significant relative to accuracy criteria assigned to the highest‐quality data in global databases. The advantages of the compact flow‐through system are presented along with a discussion of future solutions for improving data quality.