Chemiluminescence emission from exited species such as OH* or CH* as well as pressure can be convenient and effective diagnostics for monitoring ignition delay times in shock-heated mixtures. Ideally, the ignition delay time obtained from the radical-species emission signal should agree with ignition delay time as obtained from the pressure trace. Under ideal shock-tube conditions, ignition behind the reflected shock wave occurs first at the endwall, so the measurement of endwall pressure is often considered the best way to determine ignition delay time when such an increase in pressure is available. However, the signal-to-noise ratio of data from a pressure transducer mounted in the endwall can be relatively low when compared to that of an emission signal, so the latter technique provides a useful alternative to pressure. In the present paper, an analytical model of endwall emission measurements is presented, and recent experimental results are studied to determine whether or not endwall measurements are better than sidewall measurements, and vice versa. The results of this work indicate that endwall emission measurements can lead to artificially longer ignition times under dilute conditions when the increase in the radical species can occur over a pre-ignition period on the order of a hundred microseconds or more. This longer apparent ignition time is a result of the integrated effect of the detector seeing ignition occurring at later times down the length of the driven section. Endwall emission should therefore not be used to infer ignition delay times in experiments where there is no significant pressure rise and the ignition event is not abrupt. However, it is shown from experiment and the simple optics model that endwall emission can, in contrast, be employed reliably to measure ignition delay times when the ignition event is abrupt, as in undiluted fuel-air mixtures. This paper also brings to light the differences between sidewall and endwall pressure measurements in highly reactive mixtures producing strong ignition events; in such experiments, ignition delay times inferred from sidewall measurements can be artificially shorter by as much as 30 microseconds or more.