Pulse arrival-time measurements made at frequencies near 670 and 800 MHz over a 7-year period have been used to study the timing behaviour of 45 southern pulsars. These measurements have resulted in more accurate estimates of the period, period derivative, position and dispersion measure of most of these pulsars. Changes in some of the dispersion measures have been detected and used to estimate the scale-sizes and electron densities of irregularities in the interstellar medium. The timing measurements also revealed distinct jumps in the rotation rate of five pulsars, most of which have not been reported elsewhere.Polarimetric data at 670 and 800 MHz are presented for eighteen of the more luminous pulsars. The integrated profiles display properties in general agreement with trends extrapolated from other frequencies. Rotation measures have been derived for these pulsars. Nine are more accurate than previous determinations, and two have changed since they were last measured. These changes suggest a dense, magnetised electron cloud has moved into or out of the line of sight.It has been confirmed that the level of pulsar timing noise is correlated with spin-down rates. Further analyses have revealed that, in most pulsars, the observed timing noise is due to a mixture of different kinds of activity, such as random walk processes, microjumps, and for one pulsar, free precession of the spin axis. The microjumps are characterised by a variety of signatures, involving both positive and negative changes in the rotation frequency and frequency derivative. The amplitudes of these microjumps form the tail-end of a continuous distribution of amplitudes extending down to microglitch level. An investigation of timing noise in the Vela pulsar has shown that the microactivity during the postglitch recovery period is dominated by microjumps in the rotation frequency and frequency derivative which occur every 30-40 days. This microactivity is also punctuated by less frequent events, such as large steps in the frequency derivative, discrete changes in the frequency second derivative, and in one isolated case, a small glitch. The above results have been discussed in terms of current theories of pulsar timing noise and it was found that no theoretical model, in isolation, is able to explain the range of observed timing activity in the pulsars studied.Power spectra of the pulsar timing residuals have been derived using a novel technique based on the CLEAN algorithm. Most of the spectra are well described by a single-or double-component power-law model. Some of these spectra can be interpreted in the context of one or more of the current timing noise models.