Abstract Atomic carbon (C i) has been proposed to be a global tracer of the molecular gas as a substitute for CO, however, its utility remains unproven. To evaluate the suitability of C i as the tracer, we performed [C i](3P1–3P0) [hereinafter [C i](1–0)] mapping observations of the northern part of the nearby spiral galaxy M 83 with the Atacama Submillimeter Telescope Experiment (ASTE) telescope and compared the distributions of [C i](1–0) with CO lines [CO(1–0), CO(3–2), and 13CO(1–0)], H i, and infrared (IR) emission (70, 160, and 250 μm). The [C i](1–0) distribution in the central region is similar to that of the CO lines, whereas [C i](1–0) in the arm region is distributed outside the CO. We examined the dust temperature, Tdust, and dust mass surface density, Σdust, by fitting the IR continuum-spectrum distribution with a single-temperature modified blackbody. The distribution of Σdust shows a much better consistency with the integrated intensity of CO(1–0) than with that of [C i](1–0), indicating that CO(1–0) is a good tracer of the cold molecular gas. The spatial distribution of the [C i] excitation temperature, Tex, was examined using the intensity ratio of the two [C i] transitions. An appropriate Tex at the central, bar, arm, and inter-arm regions yields a constant [C]$/$[H2] abundance ratio of ∼7 × 10−5 within a range of 0.1 dex in all regions. We successfully detected weak [C i](1–0) emission, even in the inter-arm region, in addition to the central, arm, and bar regions, using spectral stacking analysis. The stacked intensity of [C i](1–0) is found to be strongly correlated with Tdust. Our results indicate that the atomic carbon is a photodissociation product of CO, and consequently, compared to CO(1–0), [C i](1–0) is less reliable in tracing the bulk of “cold” molecular gas in the galactic disk.