Born François-Marie Arouet, Voltaire (1694–1778), by the time of the Lisbon earthquake, was already a well-established author, known for his satirical wit. He had been made a member of the Académie Française in 1746. He was a deist, a strong proponent of religious freedom, and a critic of tyrannical governments. Candide became part of his large, diverse body of philosophical, political and artistic works expressing these views. More specifically, it was a model for the eighteenth- and early nineteenth-century novels called the contes philosophiques. This genre, of which Voltaire was one of the founders, included previous works of his such as Zadig and Micromegas.
It is unknown exactly when Voltaire wrote Candide, but scholars estimate that it was primarily composed in late 1758 and begun as early as 1757. Voltaire is believed to have written a portion of it while at his house in Ferney and also while visiting Charles Théodore, the Elector-Palatinate at Schwetzingen, for three weeks in the summer of 1758. Despite solid evidence for these claims, a popular legend persists that Voltaire wrote Candide in three days. This idea is probably based on a misreading of the 1885 work La Vie intime de Voltaire aux Délices et à Ferney by Lucien Perey (real name: Clara Adèle Luce Herpin) and Gaston Maugras. The evidence indicates strongly that Voltaire did not rush nor improvise Candide, but worked on it over a significant period of time, possibly even a whole year. Candide is mature and carefully developed, not impromptu, as the intentionally choppy plot and the aforementioned myth might suggest.
There is only one extant manuscript of Candide that was written before the work's 1759 publication; it was discovered in 1956 by Wade and since named the La Vallière Manuscript. It is believed to have been sent, chapter by chapter, by Voltaire to the Duke and Duchess La Vallière in the autumn of 1758. The manuscript was sold to the Bibliothèque de l'Arsenal in the late eighteenth century, where it remained undiscovered for almost two hundred years. The La Vallière Manuscript, the most original and authentic of all surviving copies of Candide, was probably dictated by Voltaire to his secretary, Wagnière, then edited directly. In addition to this manuscript, there is believed to have been another, one copied by Wagnière for the Elector Charles-Théodore, who hosted Voltaire during the summer of 1758. The existence of this copy was first postulated by Norman L. Torrey in 1929. If it exists, it remains undiscovered.
Voltaire published Candide simultaneously in five countries no later than 15 January 1759, although the exact date is uncertain. Seventeen versions of Candide from 1759, in the original French, are known today, and there has been great controversy over which is the earliest. More versions were published in other languages: Candide was translated once into Italian and thrice into English that same year. The complicated science of calculating the relative publication dates of all of the versions of Candide is described at length in Wade's article "The First Edition of Candide: A Problem of Identification". The publication process was extremely secretive, probably the "most clandestine work of the century", because of the book's obviously illicit and irreverent content. The greatest number of copies of Candide were published concurrently in Geneva by Cramer, in Amsterdam by Marc-Michel Rey, in London by Jean Nourse, and in Paris by Lambert.
Candide underwent one major revision after its initial publication, in addition to some minor ones. In 1761, a version of Candide was published that included, along with several minor changes, a major addition by Voltaire to the twenty-second chapter, a section that had been thought weak by the Duke of Vallière. The English title of this edition was Candide, or Optimism, Translated from the German of Dr. Ralph. With the additions found in the Doctor's pocket when he died at Minden, in the Year of Grace 1759. The last edition of Candide authorised by Voltaire was the one included in Cramer's 1775 compilation, l'éditions encadrées, meaning "supervised editions".
Voltaire strongly opposed the inclusion of illustrations in his works, as he stated in a 1778 letter to the writer and publisher Charles Joseph Panckoucke:
Je crois que des Estampes seraient fort inutiles. Ces colifichets n'ont jamais été admis dans les éditions de Cicéron, de Virgile et d'Horace. (I believe that these illustrations would be quite useless. These baubles have never been allowed in the works of Cicero, Virgil and Horace.)
Despite this protest, two sets of illustrations for Candide were produced by the French artist Jean-Michel Moreau le Jeune. The first version was done, at Moreau's own expense, in 1787 and included in Kehl's publication of that year, Oeuvres Complètes de Voltaire. Four images were drawn by Moreau for this edition and were engraved by Pierre-Charles Baquoy. The second version, in 1803, consisted of seven drawings by Moreau which were transposed by multiple engravers. The twentieth-century modern artist Paul Klee stated that it was while reading Candide that he discovered his own artistic style. Klee illustrated the work, and his drawings were published in a 1920 version edited by Kurt Wolff.