The frequency-dependent model for the structure of the Centaurus A VLBI source, first suggested by Preston et al. [1983] and extended by Meier et al. [1989], has been confirmed here with the first simultaneous VLBI observations at 4.8 and 8.4 GHz. These observations allow the unambiguous identification of the core of the radio galaxy, which has a highly inverted spectrum between 4.8 and 8.4 GHz. The remaining components in the sub-pc-scale jet have less inverted or flat spectra.
The first extensive monitoring of the sub-pc-scale structure of Centaurus A has been presented. The results of the monitoring show that the source consists of a number of discrete components in an underlying smooth jet. One of these components, C2, is moving with a subluminal apparent speed of approximately 0.12c, away from the core of the source along the position angle of the jet and appears to have been ejected from the core in mid 1989. Another component, C1, lies further from the core than C2 and may also be travelling with a similar subluminal speed, but episodes of strong internal evolution have been observed in C1 which complicate the interpretation in terms of a simple linear motion. It seems likely that the evolution of C1 is due to the combination of a slow linear motion and strong internal evolution on time-scales as short as a few months. The component C3 appears to have only been ejected from the core recently since it was first detected midway through the sequence of observations. The limited data collected thus far for this component shows that it may also have a subluminal speed.
High resolution and reliable images produced from a combination of data from the SHEVE array and the VLBA have resulted in one of the first detections of a sub-pc-scale counterjet. The jet to counterjet brightness ratio in conjunction with estimates of speed in the jet indicate that the jet is inclined at a large angle to our line of sight, perhaps 70 
to 80 , consistent with the kpc-scale structure of the radio source. Notably, it is difficult to reconcile the subluminal motion of component C2 with the observed jet to counterjet brightness ratio. This indicates that the motion of component C2 probably does not reflect the bulk motion of material in the jet. The episodes of rapid evolution may better reflect the true speed of material in the jet.
The spectral indices of the components seen with VLBI can be plausibly explained by the existence of a free-free absorbing structure which surrounds the central radio source and obscures the VLBI jet. The free-free absorption is in addition to synchrotron self-absorption for the compact core component and contributes to the 
spectral index observed between 4.8 and 8.4 GHz. Some information on the form of the obscuring structure can be derived only if values for the intrinsic spectral indices of C1 and C2 are assumed. The structure could plausibly be described as a torus with a cross sectional diameter of approximately 1 pc, similar to the free-free absorbing structures suggested for 3C 84 [Vermeulen, Readhead, & Backer 1994] and Cygnus A [Conway & Blanco 1995]. However, this interpretation may be somewhat complicated by single dish observations of a variable spectral index core [Botti & Abraham 1994].
Both the multi frequency and multi epoch VLBI observations of Centaurus A have provided a better understanding of the structure and evolution of a radio jet on the sub-pc-scale. A concerted program of further observations is now required to allow a more detailed study.
A key program to undertake will be the short time-scale monitoring of the sub-pc-scale structure of Centaurus A. The value of the combined SHEVE+VLBA imaging observations is clear since the counterjet is detected only when the combined u-v coverage is available. Such observations should be continued at a rate of 3 or 4 per year to follow the subluminal component motions, not only in the jet but also possibly in the counterjet. Monitoring observations should also be made with a much shorter period to monitor the short time-scale changes in the bright part of the jet, namely C1.
Another key program will be to push the angular resolution of VLBI imaging observations of Centaurus A higher. At 8.4 GHz with the SHEVE+VLBA array, the VLBI core component is still only marginally resolved with a measured size of approximately 20 light days. Since the core spectral index is highly inverted, observations at a higher frequency would be well suited for increasing the resolution of imaging observations. Such observations have recently been undertaken with the VLBA. However, the largest gain in angular resolution will come from observations with the VSOP space VLBI observatory [Hirosawa 1991] in conjunction with ground based arrays such as the SHEVE array and the VLBA. At 22 GHz, the VSOP+ground array resolution will be approximately 50 
as, corresponding to approximately one light day at Centaurus A. Extensive Space VLBI observations of Centaurus A have been accepted for phase I of the VSOP mission. Centaurus A will be a key source in one of the VSOP Key Science Programs.