The first thing to sort out is that this is not one language. Scheme is small, minimal and academic; Common Lisp is large, industrial and full of accumulated machinery. The parentheses look identical and the books are shelved together, which is why so many people start a Common Lisp book, switch to a Scheme book halfway through because someone recommended it, and end up fluent in neither. Pick a lane for the first three books at least.
The second thing is that this subject has a famous hard book at its centre, and people keep starting there. Beginning with the classic MIT text is a well-known way to quit in chapter three — not because it is badly written, but because it is a computer science course that happens to use Scheme, not an introduction to Scheme.
Learn to think in the language first
The Little Schemer is the right first book for almost everyone, in either dialect. It is a hundred and some pages of questions and answers that teach recursion by making you do it until it stops feeling strange, and it takes an evening or two. The seasoned schemer continues into continuations, state and higher-order functions, and is noticeably harder.
How to Design Programs is the gentler full-length alternative to the MIT route: a genuine introductory programming curriculum built around a design recipe, using a teaching subset of Racket. If you have never programmed before, start here rather than with the Schemer books. The Scheme programming language is the reference — Kent Dybvig's book is the one to keep open beside your editor once you are actually writing code.
The classic, at the right moment
Structure and Interpretation of Computer Programs belongs here and not earlier. Its subject is abstraction, state, streams, metalinguistic evaluation and the register machine, and it will teach you more about programming than anything else on this list, provided you do the exercises. Budget months, not weeks, and do not treat abandoning it as failure — plenty of good programmers read it in their third year rather than their first.
The Common Lisp branch
Land of Lisp is the entry point if you want the industrial dialect and want to enjoy yourself: Conrad Barski teaches Common Lisp by building games, with comics. Practical Common Lisp is the standard modern introduction and the one most working Lisp programmers recommend — free online, project-driven, and the book that covers the parts of the language you actually use.
ANSI Common Lisp covers similar ground more tersely and is the weaker choice of the two for a first pass, so read it only if Paul Graham's style suits you. On LISP is a different matter: it is about macros, and it is where the language stops being a curiosity and starts being a tool for building other languages. It has long been out of print, though the author distributes it himself. Let Over Lambda pushes macro technique considerably further; be aware that its aggressive style is genuinely contested among Common Lisp programmers, some of whom regard several of its patterns as showing off rather than good practice.
Where it leads
Paradigms of Artificial Intelligence — published as Paradigms of Artificial Intelligence Programming, and stored in our catalogue under the shortened title — is Peter Norvig's book of large worked programs, and it is the best demonstration anywhere of what an experienced Lisp programmer actually does. Lisp in Small Pieces closes the path with the deep end: writing Lisp interpreters and compilers, eleven of them, in increasing sophistication.
Two gaps worth naming. This path does not cover Clojure, which is the dialect most people encounter through work today and has its own separate shelf. And the older Common Lisp titles predate current tooling, so pair them with a modern quickstart for your implementation of choice. Our computer architecture hub is a good companion once the interpreter chapters begin.
Follow the whole path in order, pick your dialect early, and the famous hard book stops being a wall.
Follow the full ordered path here: Best Books to Learn Lisp and Scheme, in Reading Order.
