Abstract Brown dwarf spectra offer vital testbeds for our understanding of the chemical and physical processes that sculpt substellar atmospheres. Recently, atmospheric retrieval approaches have been successfully applied to low-resolution (R ∼ 100) spectra of L, T, and Y dwarfs, yielding constraints on the chemical abundances and temperature structures of these atmospheres. Medium-resolution (R ∼ 10³) spectra of brown dwarfs offer additional insight, as molecular features are more easily disentangled and the thermal structure of the upper atmosphere is better probed. We present results from a GPU-based retrieval analysis of a high signal-to-noise, medium-resolution (R ∼ 6000) FIRE spectrum from 0.85 to 2.5 μm of the T9 dwarf UGPS J072227.51–054031.2. At 60× higher spectral resolution than previous brown dwarf retrievals, a number of novel challenges arise. We examine the effect of different opacity sources, in particular for CH₄. Furthermore, we find that flaws in the data like errors from order stitching can bias our constraints. We compare these retrieval results to those for an R ∼ 100 spectrum of the same object, revealing how constraints on atmospheric abundances and temperatures improve by an order of magnitude or more with increased spectral resolution. In particular, we can constrain the abundance of H₂S, which is undetectable at lower spectral resolution. While these medium-resolution retrievals offer the potential of precise, stellar-like constraints on atmospheric abundances (∼0.02 dex), our retrieved radius is unphysically small ( R = 0.50 − 0.01 + 0.01 R _Jup), indicating shortcomings with our modeling framework. This work is an initial investigation into brown dwarf retrievals at medium spectral resolution, offering guidance for future ground-based studies and JWST observations.