Using stable isotope data obtained on multiple aliquots of diamonds from worldwide sources, it has been argued that carbon and nitrogen in diamond are decoupled. Here we re-investigate the carbon–nitrogen relationship based on the most comprehensive microbeam data set to date of stable isotopes and nitrogen concentrations in diamonds (n = 94) from a single locality. Our diamond samples, derived from two kimberlites in the Chidliak Field (NE Canada), show large variability in δ13C (− 28.4 ‰ to − 1.1‰, mode at − 5.8‰), δ15N (− 5.8 to + 18.8‰, mode at − 3.0‰) and nitrogen contents ([N]; 3800 to less than 1 at.ppm). In combination, cathodoluminescence imaging and microbeam analyses reveal that the diamonds grew from multiple fluid pulses, with at least one major hiatus documented in some samples that was associated with a resorption event and an abrupt change from low δ13C and [N] to mantle-like δ13C and high [N]. Overall, δ13C appears to be uncorrelated to δ15N and [N] on both the inter- and intra-diamond levels. Co-variations of δ15N–log[N], however, result in at least two parallel, negatively correlated linear arrays, which are also present on the level of the individual diamonds falling on these two trends. These arrays emerge from the two principal data clusters, are characterized by slightly negative and slightly positive δ15N (about − 3 and + 2‰, respectively) and variable but overall high [N].Using published values for the diamond-fluid nitrogen isotope fractionation factor and nitrogen partition coefficient, these trends are perfectly reproduced by a Rayleigh fractionation model.Overall, three key elements are identified in the formation of the diamond suite studied: (1.) a low δ13C and low [N] component that possibly is directly associated with an eclogitic diamond substrate or introduced during an early stage fluid event. (2.) Repeated influx of a variably nitrogen-rich mantle fluid (mildly negative δ13C and δ15N). (3.) In waning stages of influx, availability of the mantle-type fluid at the site of diamond growth became limited, leading to Rayleigh fractionation. These fractionation trends are clearly depicted by δ15N–[N] but are not detected when examining co-variation diagrams involving δ13C. Also on the level of individual diamonds, large (≥ 5‰) variations in δ15N are associated with δ13C values that typically are constant within analytical uncertainty. The much smaller isotope fractionation factor for carbon (considering carbonate- or methane-rich fluids as possible carbon sources) compared to nitrogen leads to an approximately one order of magnitude lower sensitivity of δ13C values to Rayleigh fractionation processes (i.e. during fractionation, a 1‰ change in δ13C is associated with a 10‰ change in δ15N). As a consequence, even minor heterogeneity in the primary isotopic composition of diamond forming carbon (e.g., due to addition of minor subducted carbon) will completely blur any possible co-variations with δ15N or [N]. We suggest this strong difference in isotope effects for C and N to be the likely cause of observations of an apparently decoupled behaviour of carbon and nitrogen isotopes in diamond.