For a queueing system with a low traffic intensity, the expected numbe
r of customers in the queue is small. However, with a bursty input pro
cess, long queues may build up in a short time. In this paper, we stud
y light traffic queueing systems with bursty input processes. We study
the distributions of queue lengths, waiting times, busy and active pe
riods, and their corresponding expansions when the traffic intensity i
s small. Special attention goes to some conditional distributions of q
ueue lengths, waiting times, and system-active periods. The expansions
given in this paper provide a potential asymptotic approach to the co
mputation of Various descriptors of queueing systems. The coefficients
of those expansions reflect some important features of episodic queue
s. We also report numerical results which give a graphic view of our a
pproximations and of the effect of the burstiness of the input and ser
vice processes on the queues.