Slaves generally speaking are highly valuable commodities because you can get such a high return on investment with them. The higher the wages in a society and the more freedom it experiences, the higher the value of the slave. The lower the population density, and the more highly regarded life is, the higher the value of the slave. The more skills the slave has and the more amicably inclined it is to being a slave, the higher its value. In a society in which slavery is socially acceptable _even to the slaves_ you are likely to see people who are technically slaves having alot of authority. The slave of a rich and powerful person, even as the least member of his household (or family), is still likely to outrank socially a poor freeman. The only exception might be racial based slavery.
Young healthy female slaves, because they can breed more slaves (and because of what that implies) will fetch higher prices than older female slaves. An older female slave is likely to be valued as a household as a nanny if the slave has a long history with the household, but is basically unsaleable (and unemployable).
Children will likely have low value, because of a high mortality rate and the fact that the return on investment is heavily delayed. Female children will out value male children.
If you care about realism, one problem you will run into HARD (like a brick wall) is the broken economics of D&D. The price of many items - especially weapons - in D&D is based of a fantasy 'gold peice standard' designed to reduce player access to goods while still giving out large hordes of pyschologically impressive treasures. However, the price of common _services_ is based off a historical 'silver peice standard'. Computing the value of slaves is going to be ridiculous under this notion, as it might well be alot cheaper to pay wages to someone than try to save money by employing and feeding a slave. This is especially true if you don't have a society which accepts slavery as basically a normal condition (which I doubt you will because the DM and players are modern people).
I suggest that the value of most slaves be intermediate in value between a palfry and a warhorse. I say this because the value of slaves is frequently compared to horses, with the value of the horse being compared favorably in most cases. Assuming you don't compute better prices for horses based on some historical research, I'd suggest that average slaves sell for between 50-150 gp depending upon age, disposition, intelligence, and gender. Low value slaves (elderly, children, hostile disposition, poor physical health, poor suitability (too intelligent for a field hand or too stupid for housework), etc.) probably should go for half of that. Very young children probably no more than 10 g.p. High value slaves (suitable for gladiatorial games, virgins with great beauty, trained courtesans, both excellent disposition and capacity to perform skilled labor (smith, carpenter, literate, etc.)) will probably fetch 2-5 times that, and will probably be treated like the valued possessions that they are.
Actual prices from modern (meaning 19th century) slave auctions are readily available. It is difficult to convert prices from fantasy gold standards to 19th century money, but at the time of auctions one 'gold peice' was worth $20. Fanstasy gold is somewhat more common place than real gold and as such one gold peice probably doesn't represent quite the wealth that it historically did. I therefore suggest that prices can be obtained by dividing the price paid for the slave by $20 then multiplying by some constant term that you feel appropriate for your campaign. If you've computed highly realistic costs of goods, multiplying by one is probably nearly correct. If you are using Player's Handbook prices, multiplying by 2 or 3 will probably give you better numbers.
I'll do some research and see if I can't come up with some numbers.
UPDATE: Well, I did some reading but it didn't clear the matter up in any way. If anything, I discovered that the price of human beings as a commodity was even more complicated of a question than I had first imagined. The only thing that I can state with certainty is that in the modern institution of slavery, the value of the slave was directly tied to the value of the commodity that their industry was expected to produce. Thus, when cotton's price was high, slave prices rose in the southern United States. When sugar prices where high, slave prices would rise in Brazil, and when coffee prices where high, the high prices paid for slaves would encourage the transportation of slaves from South America and the Caribean to coffee producing regions in Central America. Thus, we see the price paid for fields hands varying by as much as 100% over a 10 or 20 year period.
Female slaves were valued at about 80% of male slaves, which somewhat surprised me until I considered in how many societies daughters were valued much less than sons. Clearly short term economic gain outweighed the long term benifits of reproduction (perhaps because of high mortality rates during child birth)? Children were more valuable than I would have credited, perhaps a third to a half of the price of adults. Presumably, by the time the child had reached 6-8 its labor was worth something and the highest risk of childhood mortality had passed.
However, the expectation of skilled slaves fetching 2-5 times that of ordinary field hands did prove to be true.
Other than that though, I confess to not having any better of an idea what to price slaves as in D&D terms than I did with my first guess.
