I am currently an expatriate in France, and have also lived in the United States, Mexico, Spain. Hofstede's book is a good guide to better understand culture. As with a map, we recognize that it is not (nor is it intended to be reality) rather a tool to help guide us. The information on cultures in this book is our "first best guess" to understanding business norms in that culture, and then once we get to know the individuals we are working with, we can adapt. We recognize that these "norms" may change depending on the industry, the region, sub-culture, or other various factors. This book is extremely helpful in creating our "first best guess."

Hofstede's work was and is not really helpful; perhaps fascinating for those who deal with the issues from the comfortable space of their warm home of office in a western country.
For all who want to get a fundamental insight to understanding espec. Asian behavior and cultural differences, I highly recommend to start with: "The Geography of Thought : How Asians and Westerners Think Differently...and Why -- by Richard Nisbett".
That's the stuff that will bring you where Hofstede could never dream to be.

Although many comments have already been accumulated let me add something, since some of the reviews tend to get out of focus. Hofstede never claimed to have studied cultures in general, he studied effects of culture on work-related values. For this topic his work is still the standard. The starting point is like this: a large company like IBM tries to establish a strong corporate identity shared among all of its worldwide employees ("We are IBM" kind of thing). However, if you ask them a couple of questions about their work-related values, they answer differently. Turns out, the differences can be explained to some degree by the employee's country-of-origin, that is his or her culture. Hofstede then goes on and tries to find dimensions in order to describe the differences between cultures, - and it has to be said again and again - dimensions for "work-related values" and not for culture in general! This observation was and is tremendously important for multinational companies. It means that we are still influenced even when we work at a multinational firm by our cultural traditions and that this cannot easily be exchanged by the company's culture. Of course if you are more interested in other aspects of culture, than Hofstede's books might not the prime choice for you to study.
Hofstede's work is scientifically sound. The choice of IBM as a case is reasonable given his prime motivation. Sample sizes are impressive for all who have tried similar studies (besides, representativity is not a function of sample size but given by the radomness of the sample draw. Sample size has an effect on standard error but this can be taken into account with a test of significance). Quackery is how other people have used Hofstede's data in contexts other than work-related.