The dynamics of forest-atmosphere CO2 carbon isotope exchange were examined in a coniferous forest in Colorado, United States. Tunable diode laser absorption spectrometry provided extensive characterization of the carbon isotope content (δ13C) of CO2. Observed patterns in δ13C of forest air were associated with photosynthesis, respiration, and atmospheric boundary layer dynamics. Similar relationships between δ13C and CO2 were observed at all forest heights and confined to a relatively narrow envelope. Substantial variation was observed in the isotope ratio of nocturnal ecosystem respiration (δ13CR, calculated from isotopic mixing lines). A systematic bias was identified when estimating δ13CR from data sets with small range in CO2 in the samples, leading us to restrict analysis of δ13CR to periods with CO2 range >40 μmol mol−1. Values of δ13CR varied from −28.1 to −25.2‰, with variation from one night to the next as large as 1.7‰. A consistent difference was observed between δ13CR calculated near the forest floor (<2 m height) versus the upper canopy (5–11 m) on the same nights. δ13CR was more enriched in the upper canopy than near the ground on 34 of 43 nights, with a mean enrichment of 0.6‰ and a maximum of 2.3‰. A similar pattern was observed comparing δ13CR at night with the analogous quantity calculated during daytime, but only a few daytime periods met the 40 μmol mol−1 criterion. Comparisons between air samples measured (1) 10 m above the forest canopy, (2) 3 km away, and (3) within the convective boundary layer 125 km distant showed CO2 differences between sites as large as 5–6 μmol mol−1 even at midday. These results suggest that attempts to use flask measurements at remote monitoring stations as a proxy for the air directly interacting with a vegetation canopy should be made with caution. However, our results also suggest that substantial information about biosphere-atmosphere isotopic exchange can be obtained by simultaneous examination of CO2and δ13C at multiple spatial scales.