Beginning with version 3.9.0 (2015-10-14) SQLite uses semantic versioning. Prior to that time, SQLite employed a version identifier that contained between two and four numbers.
All SQLite releases starting with 3.9.0 use a three-number "semantic version" of the form X.Y.Z. The first number X is only increased when there is a change that breaks backward compatibility. The current value for X is 3, and the SQLite developers plan to support the current SQLite database file format, SQL syntax, and C interface through at least the year 2050. Hence, one can expect that all future versions of SQLite for the next several decades will begin with "3.".
The second number Y is incremented for any change that breaks forward compatibility by adding new features. Most future SQLite releases are expected to increment the second number Y. The Z is reset to zero whenever Y is increased.
The third number Z is incremented for releases consisting of only small changes that implement performance enhancements and/or bug fixes.
The rate of enhancement for SQLite over the previous five years (2010-2015) is approximately 6 increments of Y per year. The numbering format used by for SQLITE_VERSION_NUMBER and sqlite3_libversion_number() allows versions up to 3.999.999, which is more than enough for the planned end-of-support date for SQLite in 2050. However, the current tarball naming conventions only reserve two digits for the Y and so the naming format for downloads will need to be revised in about 2030.
This historical version numbering system used a two-, three-, or four-number version: W.X, W.X.Y, or W.X.Y.Z. W was the file format: 1 or 2 or 3. X was the major version. Y was the minor version. Z was used only for patch releases to fix bugs.
There have been three historical file formats for SQLite. SQLite 1.0 through 1.0.32 used the gdbm library as its storage engine. SQLite 2.0.0 through 2.8.17 used a custom b-tree storage engine that supported only text keys and data. All modern versions of SQLite (3.0.0 to present) use a b-tree storage engine that has full support for binary data and Unicode.
This major version number X was historically incremented only for large and important changes to the code. What constituted "large and important" was subjective. The 3.6.23 to 3.7.0 change was a result of adding support for WAL mode. The 3.7.17 to 3.8.0 change was a result of rewrite known as the next generation query planner.
The minor version number Y was historically incremented for new features and/or new interfaces that did not significantly change the structure of the code. The addition of common table expressions, partial indexes, and indexes on expressions are all examples of "minor" changes. Again, the distinction between "major" and "minor" is subjective.
The patch level Z was historically only used for bug-fix releases that changed only a small number of code lines.