Translate your topic into English. Look for alternative search words and synonyms for your main concepts. Use for example dictionaries, thesauri, index terms (subject headings), reference books, course books, or Wikipedia to find suitable words and phrases. If you have already found a good source, take advantage of the bibliography and the words and phrases used in it. Check if your topic is clear and unambigous, or if the terminology differs for example depending on the continent it is used in. Also look for spelling variations (GB-eng/US-eng), abbreviations, as well as Latin names and terms.
Search for MeSH terms (also called subject headings or index terms) for your topic in PubMed. Look for good articles in PubMed and check what MeSH terms have been used to describe them. Also go through the other search words and phrases you have found to see if any of them have been used as MeSH terms in PubMed. Add the terms to your search word list.
Form a search string from the words, phrases, and index terms and run a test search. If it looks like some of the search words are bringing inaccurate search results, consider replacing/removing them. If PubMed's search results are very different from what you expect, check the search history in Search Details to see what is happening with your search. You can bypass PubMed's automatic term mapping by focusing your search words to Text Word fields by using a [tw] tag at the end of the words. Add relevant search words you find along the way to your search string, run the search, and use search filters if necessary.
Repeat steps 1-3 as needed until you have created a working search that brings a sufficient amount of relevant search results. Note that a search string is not supposed to be so narrow that it only finds results relevant to your topic; you always have to go through the results and pick the relevant ones.
Compose a search string to each database you will use. Different databases have different index terms (subject headings) or no index terms at all. Use of truncation and phrase marks can also vary depending on the database. Be precise with Boolean operators and parentheses in the search string.
Once you are happy with the search strings you are planning to use in different databases, make sure they fulfil the requirements for a systematic search strategy (publication date, language, or field tag filtering; used search words, etc.).
Double-check that your search string is composed following the search rules of the database.
Locate the fulltext resources you have found with your searches. In most of the databases a Volter logo will be visible in the search results. You can use it to check if the university library has electronic access to the articles. With older resources you may be able to use interlibrary services and order article pdf files for free from the National Repository Library in Kuopio. A growing number of articles can also be found on the internet as Open Access versions free to read by everyone. The parallel publications are often manuscript or pre-print versions that have been made available after the publisher's embargo period has lapsed.