We present a computer library for the numerical evaluation of colour-ordered n-gluon amplitudes at one-loop order in pure Yang–Mills theory. The library uses the recently developed technique of generalised unitarity. Running in double precision the library yields reliable results for up to 14 gluons with only a small fraction of events requiring a re-evaluation using extended floating point arithmetic. We believe that the library presented here provides an important contribution to future LHC phenomenology. The program may also prove useful in cross checking results obtained by other methods. In addition, the code provides a sample implementation which may serve as a starting point for further developments. Program summaryProgram title:NGluonCatalogue identifier: AEIZ_v1_0Program summary URL:http://cpc.cs.qub.ac.uk/summaries/AEIZ_v1_0.htmlProgram obtainable from: CPC Program Library, Queenʼs University, Belfast, N. IrelandLicensing provisions: GNU Public LicenseNo. of lines in distributed program, including test data, etc.: 30 677No. of bytes in distributed program, including test data, etc.: 334 896Distribution format: tar.gzProgramming language: C++Computer: Any computer platform supported by the GNU compiler suite.Operating system: No specific requirements – tested on Scientific Linux 5.2.RAM: Depending on the complexity, for realistic applications like 10 gluon production in double precision below 10 MB.Classification: 11.5External routines: QCDLoop (http://qcdloop.fnal.gov/), qd (http://crd.lbl.gov/~dhbailey/mpdist/)Nature of problem: Evaluation of next-to-leading order corrections for gluon scattering amplitudes in pure gauge theory.Solution method: Purely numerical approach based on tree amplitudes obtained via Berends–Giele recursion combined with unitarity method.Restrictions: Running in double precision the number of gluons should not exceed 14.Running time: Depending on the number of external gluons between less than a millisecond (4 gluons) up to a 1 s (14 gluons) per phase space point.