EAZY is a fast, public photometric redshift code designed for cases where spectroscopic redshifts are unavailable or only available for a biased subset of galaxies. It combines features from existing codes, including linear combinations of templates, optional flux- and redshift-based priors, and a user interface modeled after HYPERZ. The code uses semi-analytical models for templates and priors, rather than spectroscopic samples, and includes a novel rest-frame template error function to account for wavelength-dependent template mismatches. A redshift quality parameter, Q_z, is introduced to estimate the reliability of photometric redshifts. EAZY performs well on public datasets, achieving a 1σ scatter of 0.034 in Δz/(1+z) for K-selected samples in CDF-South and other deep fields. It provides updated photometric redshift catalogs for the FIRES, MUSYC, and FIREWORKS surveys. The code uses a nonnegative matrix factorization algorithm to derive an optimized template set, which is designed for deep optical-NIR surveys and does not require optimization based on spectroscopic samples. The template error function is derived from the GOODS-CDFS photometric catalog and accounts for uncertainties in template fitting. Bayesian priors are used to improve redshift estimates, with the prior probability distribution based on the redshift distribution of galaxies in the lightcone catalog. The code produces two redshift estimates, z_p and z_mp, and computes confidence intervals for the redshifts. EAZY is fast and efficient, with a parameter file that provides default inputs for most applications. It has been tested on a combined photometric catalog and compared to a large sample of spectroscopic redshifts, showing good agreement with spectroscopic redshifts. The code's performance is compared to other photometric redshift codes, including neural network-based methods, and it is found to have lower scatter and systematic errors than some other methods. The reliability of photometric redshifts is assessed using confidence intervals and a reliability parameter, Q_z, which provides a robust estimate of the reliability of the photometric redshift estimate.EAZY is a fast, public photometric redshift code designed for cases where spectroscopic redshifts are unavailable or only available for a biased subset of galaxies. It combines features from existing codes, including linear combinations of templates, optional flux- and redshift-based priors, and a user interface modeled after HYPERZ. The code uses semi-analytical models for templates and priors, rather than spectroscopic samples, and includes a novel rest-frame template error function to account for wavelength-dependent template mismatches. A redshift quality parameter, Q_z, is introduced to estimate the reliability of photometric redshifts. EAZY performs well on public datasets, achieving a 1σ scatter of 0.034 in Δz/(1+z) for K-selected samples in CDF-South and other deep fields. It provides updated photometric redshift catalogs for the FIRES, MUSYC, and FIREWORKS surveys. The code uses a nonnegative matrix factorization algorithm to derive an optimized template set, which is designed for deep optical-NIR surveys and does not require optimization based on spectroscopic samples. The template error function is derived from the GOODS-CDFS photometric catalog and accounts for uncertainties in template fitting. Bayesian priors are used to improve redshift estimates, with the prior probability distribution based on the redshift distribution of galaxies in the lightcone catalog. The code produces two redshift estimates, z_p and z_mp, and computes confidence intervals for the redshifts. EAZY is fast and efficient, with a parameter file that provides default inputs for most applications. It has been tested on a combined photometric catalog and compared to a large sample of spectroscopic redshifts, showing good agreement with spectroscopic redshifts. The code's performance is compared to other photometric redshift codes, including neural network-based methods, and it is found to have lower scatter and systematic errors than some other methods. The reliability of photometric redshifts is assessed using confidence intervals and a reliability parameter, Q_z, which provides a robust estimate of the reliability of the photometric redshift estimate.