Freudian psychology (and it's derivatives) are indeed pseudoscience, by and large. That said, Freud was arguably the first to systematically theorize and study human behavior and cognition, and in so doing laid the foundation for the scientific study of psychology. It's generally useful for the purpose of such discussions to distinguish between Freudian psychoanalysis and Freudian psychodynamics. The latter is generally sound though embryonic to the point of near-uselessness to contemporary students (except of course, from a historical perspective). As such, there are concepts in Freud's work that have been scientifically validated.
The basic premise that we're unaware of the majority of our mental processes, for example, is perfectly valid. More generally, the idea that behavior is the sum of concrete biological forces (what Freud called psychic energy) has been substantiated by quantitative evidence time and time again. Once again, the problem is that it's very superficial and hand-wavy by modern standards, not to mention its propensity to be severely abused.
Modern attachment theory, that is, the idea that emotional bonds are an essential part of infant development, also very clearly originates with Freud.
Most interestingly, IMHO, is the concept of defense mechanisms. Indeed the study of psychology and cognitive science has been impeded for years because of an over-reliance on subjective self-report. Thanks to Freud, we recognize that people can be unaware of their own misrepresentations of fact.
It's actually quite difficult to find a nook of cognitive science that hasn't been positively influenced by Freudian psychology in some meaningful way. Freud got many things right, but wasn't very concerned with establishing the validity of his claims. In practical terms, this means that studying Freud from a historical perspective will be quite informative, but one will have a hard time applying his theories directly to modern research or therapy. And of course, there are a significant number of Freudian claims that have been positively disproven in all but the most metaphorical sense (the general "death wish" comes readily to mind).
Another way to look at this is as a modeling problem. Freudian psychology proposes a number of models of how the human psyche works, most notably the Id/Ego/Superego hierarchy. At the time, this was a very useful model, but it has since been supplanted by better models that better predict behavior. This having been said, Freud's groundwork is readily visible: everybody agrees that automatic non-conscious processing occurs and that it influences conscious perception. Similarly, it's well known that top-down influences of expectation, cultural bias or attention can affect conscious processing. In this sense, the Id/Ego/Superego model is alive and well... it's just been refined a great deal.
