Preface: For the statistical purpose of this answer, assume that all test-takers are completely ignorant of the subject material, and are all selecting answers at random. Obviously, studying for a test will skew your results toward the positive end of things, but that's not applicable to your question.
Okay. When someone says there's a 50/50 chance of something, it means that there's a 50% chance that it will happen, and a 50% chance that it will not. Your 50/50 chance of getting a T/F question correct on a test is based on the assumption that the actual answers are split evenly between true and false in the test. This is also known as a uniform distribution, in which each possible outcome occurs with equal frequency in the results. On a test, you have no knowledge about the distribution of answers, so since there's no reason to prefer one choice over another you can usually assume that you're working with a uniform distribution. It's not often a correct assumption, but if you're guessing at answers on a test, it's the best you can do.
If you have a uniform distribution of T/F on a test, then random guessing should get you a 50% on the test. Answering all T or all F should likewise get you a 50% on the test. There's no reason to prefer any answer over another if you're assuming the answers are distributed on a uniform distribution; any answer is just as good as any other.
On the other hand, imagine that you know that the professor, a horrible dart-thrower, assigned answers by whether or not he hit the bulls-eye on his dart board. In this case, you could assume that most of the answers on the test should be F, not T. In this case, if you stick with your assumption of a uniform distribution, you do not
have a 50/50 chance of getting the answer correct. A 50/50 chance implies that each choice has a 50% chance of being correct. In this test, the Ts have a much smaller chance of being correct. The probability of each choice is dependent not only on the number of choices that exist, but the distribution between them.
To take a more extreme example, note that there are two possibilities for tomorrow:
1) A massive asteroid will hit the Earth.
2) A massive asteroid will not hit the Earth.
(Assume for our purposes that anything other than actual contact counts as a miss, no matter how near the miss.) There are only two choices here, but I don't think that anyone would say that there's a 50/50 chance either way. Claiming a 50/50 chance on such conditions would imply that just as many people would be right betting against the Earth as would be right betting for it. I'm quite confident that the odds are strongly against tomorrow's asteroid collision.
So really, a 50/50 chance only is relevant in two situations. The first is when you know that two options are both equally likely, such as in tossing a coin and predicting which face it will land on. The second is when you have no information about the relative probabilities between two options, so you just assume that they're equally likely for the purposes of having a basic assumption. The second case is actually somewhat of a misnomer, since there is rarely a true uniform distribution across unknown data, but we use the 50/50 description anyway. In this case, it's more a description of our perception
of the probabilities than it is a description of the true distribution.
Hope that helps!