Anki vs Quizlet vs Memrise: 500 Cards, Measured

By Daniel Herrera, Spanish teacher and curriculum designer

Updated September 7, 2026 · Daniel Herrera

Every guide to Spanish flashcards opens the same way. Learn the 500 most common words and you will understand 80% of what you read. Pick the right app, put in the reps, and the books open up.

We have published that number ourselves, in our list of the 500 most common Spanish words. So we decided to check it against actual books.

We took 8 public-domain Spanish novels from Project Gutenberg, from Spain, Argentina and Peru, plus Don Quijote and a 1918 graded reader for students. That is 1,141,520 running words. Then we asked a simple question. If you knew the commonest words in Spanish prose, how much of the next page would you actually recognise?

The answer changes what a flashcard app is for.

How did we measure this?

We built the frequency list out of the novels themselves. Then we tested each novel against a list built from the other novels only. No book was ever scored against a list that had already read it. That matters. Score a book against its own word list and the number comes out flattering and meaningless.

We counted two ways, because there are two honest answers.

• Word forms. The word exactly as it sits on the page. Hablo, hablaba and hablarán are three.
• Stems. Those three collapse into one. This is the generous count, and it flatters flashcards, because a deck usually teaches one card per dictionary word. We used the standard Snowball stemmer for Spanish. A stem is an approximation of a dictionary word, not a perfect one.

One thing in our method favours the flashcards. Our frequency list is built from books. A real beginner deck is built from everyday speech. A book-derived list should predict books better than a speech-derived deck does. So the real gap for a shop-bought deck is probably wider than what we measured, not narrower.

How much of a Spanish novel do 500 words actually cover?

61.7% of the words on the page. Counted the generous way, by stems, 69.5%.

Here is the promise set against the measurement.

Look at the first row. At 100 words the claim is almost exactly right. We measured 48.9% against a promised 50%. The claim starts true and then breaks fast.

By 500 words the gap is 18.3 points. By 1,500 words it is 19.3.

Why is the gap so wide?

Spanish inflects, and English speakers underestimate how much.

Take the 500 commonest stems in our corpus. On the page they appear as 5,212 different written words. That is a mean of 10.42 forms for every one you would put on a card.

The verb pasar is the worst offender in our corpus. It turns up as 60 distinct written words: paso, pasar, pasado, pasó, pasos, pasaba, and more.

A deck that says 500 words is describing its own contents honestly. It is just not describing what happens when you open a book.

So how many words do you actually need?

This is the number worth knowing, and almost nobody publishes it. It is also the number that should shape how you build a Spanish vocabulary in the first place.

Half the page costs you almost nothing. 114 word forms, or 76 stems, and you are there. Those are the articles, the prepositions and the handful of verbs everything else hangs off.

After that it gets expensive quickly. 80% of the page costs about 4,388 forms. That is roughly 9 times the deck you were promised.

90% costs about 17,211 forms, and in most of our novels the 95% mark was never reached at all inside the book.

Which flashcard app is best for Spanish vocabulary?

Now the app question, with the scale in mind. You are not building a 500 card deck. You are building something that grows for years. So the things that matter are whether you can make your own cards easily, who controls the review schedule, and whether you can take your work with you. For everything else these apps do, see our round-up of Spanish learning apps.

Anki

Anki is free on computer, free on Android, and paid on iPhone, where the purchase funds the project. It is open source. The review scheduling is documented in its own manual and you can change it.

The part that matters for reading is the export. Anki will hand you your notes as plain text, a single deck as a file, or your entire collection in one package. Ten years of Spanish words stay yours.

It is also the ugliest of the three and the one that asks most of you on day one. That is a real cost, and it is the honest reason people bounce off it.

Quizlet

Quizlet is the easiest to start with, and it has an enormous library of sets other people have already made. If someone has studied it, there is a set for it.

The catch is where the spaced repetition sits. On Quizlet’s own upgrade page, the adaptive study modes and the scheduled review that tracks long-term memory are listed under Quizlet Plus, at $35.99 a year or $7.99 a month. Free Quizlet gives you flip-through cards. The part that makes flashcards work is the part you pay for.

Memrise

Memrise built its reputation on community courses. Users made them, and for Spanish they were the best thing on the platform.

Those courses can no longer be reached in the Memrise app. Memrise’s own help pages say so and point learners to a separate site, community-courses.memrise.com, to carry on. The company explains that it wanted to control the words you learn so they match the conversations you want to have.

That is a coherent decision. It also means the thing that made Memrise worth choosing for Spanish is no longer part of the app you download.

Do graded readers really have a smaller vocabulary?

This is where we expected an easy answer and did not get one.

A graded reader is supposed to be a smaller world. Fewer words, carefully chosen. So we cut every text in our set to the same length, 32,562 running words, and counted how many distinct stems each one used. Cutting to the same length matters. A long book meets more words simply by being long.

The graded reader used 3,874 stems. The median novel used 4,257. That is only 1.1 times more.

And Don Quijote, the book everyone treats as the summit, used 3,259. That is fewer distinct stems than the beginner reader, over the same stretch of text.

Which tells you what difficulty actually is. Don Quijote is hard. It is not hard because it uses more different words per page. It is hard because the words are four hundred years old, the sentences are long, and the jokes assume things you do not know. Vocabulary breadth is one ingredient, and it is not the main one. No deck size unlocks a book on its own.

What should you actually do with flashcards?

Stop using them as a gate you pass before reading. Start using them as a net you hold while reading. This is the same principle behind the techniques that actually work, and it is why a word met in a story outlasts a word met in a list.

What this study does not tell you

We want to be straight about the edges of this.

• These are novels, and old ones. Our texts are public domain, so they are mostly late nineteenth century. Modern journalism and modern dialogue will run easier than this.
• The 80% claim is usually made about conversation, not books. It may well hold better for speech. Our finding is that it is being sold to people who want to read, and for reading it does not hold.
• Stems are not dictionary words. The Snowball stemmer is a rule-based tool. It gets most Spanish right and some of it wrong, which is why we publish the word-form number beside it.
• We did not test the apps over months. We compared what each one charges for, what it lets you build, and what it lets you take away, from each company’s own documentation. We are not claiming to have measured which one teaches you faster, because we did not.
• Recognising a word is not knowing it. Coverage counts words you would recognise. Following a sentence needs more than that, which is also why people read Spanish far better than they hear it.

Frequently asked questions