Context. Current generation millimeter wavelength detectors suffer from scaling limits imposed by complex cryogenic readout electronics. These instruments typically employ multiplexing ratios well below a hundred. To achieve multiplexing ratios greater than a thousand, it is imperative to investigate technologies that intrinsically incorporate strong multiplexing. One possible solution is the kinetic inductance detector (KID). To assess the potential of this nascent technology, a prototype instrument optimized for the 2 mm atmospheric window was constructed. Known as the Néel IRAM KID Array (NIKA), it has recently been tested at the Institute for Millimetric Radio Astronomy (IRAM) 30-m telescope at Pico Veleta, Spain. Aims. There were four principle research objectives: to determine the practicality of developing a giant array instrument based on KIDs, to measure current in-situ pixel sensitivities, to identify limiting noise sources, and to image both calibration and scientifically-relevant astronomical sources. Methods. The detectors consisted of arrays of high-quality superconducting resonators electromagnetically coupled to a transmission line and operated at ~100 mK. The impedance of the resonators was modulated by incident radiation; two separate arrays were tested to evaluate the efficiency of two unique optical-coupling strategies. The first array consisted of lumped element kinetic inductance detectors (LEKIDs), which have a fully planar design properly shaped to enable direct absorbtion. The second array consisted of antenna-coupled KIDs with individual sapphire microlenses aligned with planar slot antennas. Both detectors utilized a single transmission line along with suitable room-temperature digital electronics for continuous readout. Results. NIKA was successfully tested in October 2009, performing in line with expectations. The measurement resulted in the imaging of a number of sources, including planets, quasars, and galaxies. The images for Mars, radio star MWC349, quasar 3C345, and galaxy M 87 are presented. From these results, the optical NEP was calculated to be around 1 × 10-15 W/Hz1/2. A factor of 10 improvement is expected to be readily feasible by improvements in the detector materials and reduction of performance-degrading spurious radiation.