Actually they were surprisingly high, depending on the period you are talking about. Once the army became professional it was certainly in their generals best interest to see that they were well fed and reasonably cared for. This insulated them, to varying degrees from such things as famine and malnutrition in general.

We know that promises made to large numbers of returning veterans could actually be a problem for Rome, or rather for the senate. Often times these men were promised land at the end of their service, this acted as a sort of pension. Pompey himself butted heads with the senate over this when he returned from campaigning in the east and requested the land that had been promised to his thousands of returning veterans. The senate balked at this. Land reform had been a huge deal for many years at this point and the senate class, mostly large estate holders were still disinclined to relinquish any of it, even for veterans. There were some reforms, towards the end of the republic I believe, which were intended to limit the size of an estate so there could be more land available for the veterans, and other citizens, but these basically didn't work as the landowners found loopholes and ways around the law.

As for being killed in battle there are a couple of things to keep in mind. Rome, for hundreds years was the biggest, baddest, dog on the block. Certainly they suffered defeats, some staggering defeats, but these were few enough that they stand out in history. Rome won most of its battles. In ancient battles the vast majority of deaths did not take place during the initial fighting but rather when one side broke and ran. They were then generally simply cut down by the other's Calvary. Roman legions rarely broke. And on those occasions where they did have to retreat they usually had a strong fall back position. The Romans were meticulous about building fortified camps. They build what were essentially forts for each camp. If the army was forced to retreat they fell back to this fortified position and it was usually enough to buy them time to regroup.

So given that Roman soldiers usually won their battles, and that they were often better fed than the general population, and given that we know large numbers of veterans was often a problem back Rome, I would say, yes, the odds of surviving your service were actually pretty good.

I am in the process of moving at the moment so my copy is current packed, but the book by ROME AT WAR: FARMS, FAMILIES, AND DEATH IN THE MIDDLE REPUBLIC by Nathan Rosenstein put death rate north of 30% of young men, who enter the army, serve for about 10 years, come home, take up farming, and get married. The death toll is a combination of those killed in action, and the underlying mortality rate for the age. By the time we get to the early empire, when Augustus fixes the length of service, is only a couple of years long than what had been the prevailing tradition. There is a fair amount of guess work in this book, but it comes complete with error bars, and justifications for the initial estimates.

The book itself is fantastic if you like that sort of thing.