Deciphering Japanese sword signatures (mei) is an extremely difficult business requiring much study and hard work. If you dislike complexity, if you have a short attention-span or if you like your 'browser' because you like browsing with it, turn back now before it is too late, or you may be doomed to a life of inexorable madness and confusion.
You have been warned!
If any money depends on the information you hope to find here (e.g. if you are about to buy a Japanese sword and are trying to read the signature beforehand), forget about it and go and consult a recognised expert. Any of the Japanese Sword Societies will be pleased to help you avoid wasting your cash.
Signatures are the LAST thing to look at when assessing a Nihontô, and are only viewed to verify what should have been been discovered by looking at other aspects of the blade.
To confuse matters there are tens of thousands of swords with fake signatures (gimei) which have been produced for centuries, either to deceive (usually bad news for you) or as a mark of respect to some master smith (could be good or bad news). Then there are swords produced by guilds, where many smiths would use the same standard signature of the founder of the guild or school (often quite good news).
Still here? Well, after all the caveats, you will be glad to learn that signatures (mei) usually appear on the majority of Japanese sword tangs (nakago) in a traditional, stylised fashion. This makes it easier to decipher many mei knowing little or no Japanese. It is this 'standard' form I am concentrating on here. There are many exceptions (see Varieties of Mei) which are beyond the scope of this short document however, and in these cases it is usually necessary to have a knowledge of the Japanese language and/or extensive experience of Nihontô to make an accurate translation. So if you come across a signature which does not match up with what I am saying here, consult a recognised expert. Be aware that there are people around who will claim to be able to tell you all about your Japanese sword because they once saw one in an antique shop window and saw The Seven Samurai thirty years ago!
The first task then is identifying the mei on the blade. Always hold the Nihontô point uppermost to view inscriptions. The mei is always written on the nakago such that it faces outwards (away from the body) when the sword is being worn. As a katana and a wakizashi (short-sword) is always worn edge up, when you are viewing the inscription the mei is towards you when the sharp edge (ha) is to your left. If the blade is a tachi (long sword) which is worn edge down, then when you are viewing the mei is visible when the sharp edge is to your right.
Quite often a blade will have an inscription on both sides of the nakago. In many cases the second inscription is a date, so it will pay you to learn the kanji which represent some of the numbers in Japanese so you are able to recognise this.
More detail is available about reading dates using reign names ( Nengô ), those using the Zodiacal Method and those calculated from the reign of Emperor Jimmu
Once you have located the mei, on the majority of blades you will see four, five or six Japanese characters (kanji). They read from top to bottom and they are traditionally arranged so that the first and second kanji represent the province where the smith worked/lived.
Each province is represented by two kanji.
Examples are: BIZEN 
and YAMASHIRO 
The second character is often SHÛ 
together with one other kanji from the province name. This is used as an abbreviated form. The abbreviated form of BIZEN is BISHÛ 
and that of YAMASHIRO is SHIROSHÛ 
(pronounced JÔSHÛ)
A complete list of 'Roads' and Province Names and abbreviations is now available.
The last two kanji in the mei are usually the name of the smith (note however, that many smiths used 3, 4, 5 and even 6 kanji in their signature).
Occasionally the last characters in the mei might be SAKU 
(made), SAKU KORE 
(made this), KITAU 
(forged) or TSUKU(RU) 
(made). These may be accompanied by KIN 
(respectfully).
In these cases the name of the smith is usually the two
characters above this (although to complicate matters even further the
second character of a few smiths' names is SAKU, AND there are names
which use the kanji with different pronunciations such as NARI!)
A list of the kanji used in the given names of swordsmiths is now available in romaji order : A to H / I to N / O to Z
A few smiths used mei consisting wholly or partly of katakana or hiragana characters. A list of these smiths is available.
The remaining central kanji in the mei inscription might refer to a Date (see links above), a Feudal Rank, Tameshigiri (cutting test), or consist of kanji such as (NO) KUNI 
(province), OITE 
(at) or (NO) JÛ 
(dwelling in). The presence of one of these last two may imply that the smith was not a native of that particular province.
A mélange of Michis
The major problem encountered in trying to decipher mei is that kanji are often distorted as in michi above. Take your own name, see it in print, then handwritten and then signed by you. If you compare all these you can begin to appreciate the range of distortion encountered when reading the Japanese characters in mei. There are no short-cuts here, and all serious Nihontô scholars make it their business to study oshigata or tang-rubbings of blades made by known smiths. In this way they gradually develop an 'eye' for reading and recognising kanji. The journals of the various Nihontô Societies provide a steady, affordable supply of oshigata for study and practice.
A useful way to identify an individual kanji in the signture is by using the kanji lookup software in a Japanese word processor. By this means you can obtain the JIS (Japanese International Standard) numbers of the kanji with a minimum of effort, enabling you to write about them, use them in print or communicate them to others (e.g. on forums and mailing lists) either by quoting the JIS numbers or directly by displaying them in e-mail.