A full-length classic HSK 4 paper with the official shape: 100 questions in 105 minutes across listening (45), reading (40) and writing (15). It is original practice material written for Hanlu, not an official or past paper.
Starting a paper needs an account, so the timer, autosave and your gap report can be stored against you. Create account · Log in The first section is included on the free plan.
This is unofficial practice material written for HSK Guide. It is not a past paper and is not endorsed by any examination authority.The section lengths, the timings and the scoring below are HSK Guide’s own practice settings, chosen so that a wrong answer points at one identifiable skill. They do not describe any examination authority’s rules — confirm those with your test centre.
Sections
3 sections, 100 questions, 95 minutes of section time.
Forty-five questions in three parts. Part 1 (1–10): judge the ★ statement against the recording. Part 2 (11–25): short dialogues. Part 3 (26–45): longer dialogues, then passages that each carry two questions. Each recording plays once.
Audio rules for this section
Forty questions in three parts. Part 1 (46–55): fill each gap from a six-word list. Part 2 (56–65): put three sentences in order. Part 3 (66–85): read each passage and answer the question; the last three passages carry two questions each.
Fifteen questions in two parts. Part 1 (86–95): put the words in order to make one sentence. Part 2 (96–100): look at the picture and write one sentence that uses the word given. Part 2 accepts the listed model sentences; the rubric in each explanation shows what a marker looks for.
Questions, options and answers are not shown on this page — the paper is only meaningful if you meet it for the first time inside the timer.
Before you start
The timer runs on the server, so refreshing will not give you extra time. Listening audio plays a fixed number of times, as set for this practice paper.
Total: 105 minutes · section time adds up to 95 min.
No timer. Audio can be replayed freely and answers are explained immediately.
The play count and the seek and pause rules are stored on each section rather than hard-coded in the player, so an editor can change them without a deployment.
Your result
A score is the least useful part of a mock examination. The report is built to change what you study tomorrow.
A score per section and per skill, so a listening problem cannot hide behind a good reading result.
Accuracy broken down by question type — matching, cloze, ordering, true/false — because they fail for different reasons.
The exact words the wrong answers tested, taken from the diagnostic tags on each question.
The patterns behind the wrong answers, each linking straight to its grammar lesson.
Where you stalled. A slow correct answer is a different problem from a fast wrong one.
Add all gaps to my review queue — one click adds every gap the report found to your review queue, scheduled by the same engine that runs your daily session.