We report here the results of a comparative theoretical study of ultrasonic echo imaging systems using pulsed and pseudorandom sources. We first point out the fact that linear imaging systems are characterized by two parameters: i) the resolution and ii) the signal-to-noise ratio in the image. While the maximum resolution attainable in an echo imaging system is limited by the bandwidth of the transducer(s) used, the signal-to-noise ratio depends both upon the nature of the electrical signals applied to the transducer and the type of processing performed on the echos. In this paper we show that for a given transducer and a given backscattering element located at a specified distance from the transducer one obtains a higher signal-to-noise ratio with pseudorandom signals than with pulsed signal only if the following condition is satisfied BT pr >2π 2 V pbd V cbd where B, V pbd and V cbd are the bandwidth, the peak-breakdown voltage and the cw breakdown voltage, respectively, of the transducer, and T pr is the length of the pseudorandom signal used. In this paper we also show that, if no constraints are placed on the thickness of the object, the theoretically derived signal-to-noise ratios possible with pseudorandom sources are practically realizable only when the echos from different backscattering centers within the object are non-overlapping , and each echo is circularly correlated with the transmitted signal. If the echos are overlapping, the optimum signal-to-noise ratios are possible only if the thickness of the object is less than ( cT pr 2 , where c is the velocity of propagation in the object. Also, now each burst of the transmitted signal must consist of two periods of the pseudorandom signal.