Great answers so far. I'll suggest another approach similar to "snowball sampling" in social science.
Basically, you let really good/important articles (and journals) lead you to other really good/important articles (and journals).
Start with a few articles that stand out to you (or your adviser or mentor). They might be survey articles that assess the state of research and future directions, or they might be seminal articles that spawned new lines of research.
Then look at the editorial policies and editorial boards for the journals that published these papers. Are there any patterns that seem to tie these particular journals to these particular articles?
Now look at the articles that are cited in these initial papers, especially those that are described as "seminal" or "pioneering", etc. Repeat the process of looking at the journal editorial policies and editorial boards for these articles.
Also, using Google Scholar, look at the articles that cite your original set. Which journals seemed to publish the most follow-up articles, and promote debate or contrary research? Were there Special Issues of journals devote to the ideas or methods in the original set?
Rinse and repeat :-).
Through this procedure, you will usually settle on a set of 2 to 5 journals in a field (or sub-field) that are considered "important" for the type of research you are most interested in.
The advantage of this method is that you will home in on the journals that are important for your interests and focus, not just what everyone else (worldwide) might think is important.