David has an extensive summary of the trade offs. While technology has advanced in the intervening three decades (yes, MSR has been seriously studied for longer than that), the physics remain unchanged, and, as so far as I know, the basics of rocket fuels has not advanced much.
To cut to the chase, direct return to Earth would require too much mass to be launched to Mars for it to be practical.
To give an idea of how difficult MSR is, think of all the steps that much be successfully accomplished:
Find an excellent place to collect samples. Remote sensing from orbit helps, but you can get skunked. As a thought exercise, several Mars researchers looked at various outcomes from the Mars Science Laboratory, and here's their description of one hypothetical scenario for getting skunked at what looked from orbit as a promising site: "
MSL goes to the delta deposits of Eberswalde. The delta depositional environment is confirmed, but we discover that this kind of facies on Mars is not good for preservation of organic compounds and evidence of fossil life. The clay minerals are limited to thin rock coatings, with minimal amounts of water involved in their formation. There is no sign of organic compounds, likely due to the oxidant compounds found in the rocks. "
A rover must be dispatched and safely landed on Mars to collect the samples. This rover could either arrive separately from the ascent vehicle lander (i.e., some years before in a previous launch window) or with the ascent vehicle. In the former case, we would know that interesting samples were found. However, even in this case, the lander with ascent vehicle would need to carry a small rover to go fetch the sample cache from the previously flown rover. (MSR could be done without a rover, but think of all the uninteresting locations the MER rovers have visited to find the really interesting locations. Imagine if a MSR lander without a rover had landed in the same place as Opportunity. It could stare at that wonderful bedrock just a few meters away but out of reach without a rover.) The ascent vehicle plus a rover and other sampling tools much be safely landed on Mars. Samples must be transferred from the rover(s) to the ascent vehicle. The ascent vehicle must then launch and reach Martian orbit. The Earth return orbiter must enter Martian orbit and autonomously rendezvous with the ascent vehicle. The sample cache must be transfered from one vehicle to another. The Earth return orbiter must break free of Martian orbit and carry the samples back to Earth. At Earth, the samples are returned in an atmospheric return capsule much like what was done for the Stardust comet samples.
MSR promises to provide a scientific bonanza if all these elements (many of which require extensive technology development) can be made to work. If any element fails, then the entire investment is lost. This is why estimates for MSR range from $3B (seems too low in my amateur armchair estimation) to $5-6B. Fortunately, all the separate elements makes international cooperation fairly straightforward. That, too, could be problematic. If any nation withdraws its development of a key element, then the investments of the other nations are put at risk.
A lot of momentum has built behind MSR in the last couple of years. It will be interesting to see if the issues above are addressed successfully by the international consortium of space agencies that have expressed interest.
Science priorities for MSR