A dependency in project management is a rule that ties two tasks together: one task can’t start, or can’t finish, until another task has started or finished. You can’t pour concrete until the foundation is dug, so pouring the concrete depends on digging the foundation.

The task that comes first is called the predecessor. The task that waits is called the successor. Every task has a start and a finish, and a dependency joins one end of the predecessor to one end of the successor. There are only four ways to join two ends, so there are four types of dependency.
Part 1: The four types
There are four types of dependency in project management
In the pictures below, the predecessor is the blue bar and the successor is the purple one. The arrow shows which end waits for which.
| Type | Everyday example | Work example |
|---|---|---|
| Finish to start (FS) | Eat the cake after it has finished baking | Pour the concrete after the foundation is dug |
| Start to start (SS) | The dancing starts once the music starts | Pipe laying starts once trench digging starts |
| Finish to finish (FF) | Lecture notes finish when the lecture does | Testing finishes when development does |
| Start to finish (SF) | The night guard goes home when the morning guard arrives | The hired generator goes back when mains power is switched on |
Finish to start: the second task waits for the first to finish
This is the most common type by far, and the one most planning software uses when you link two tasks without choosing. You can’t eat a cake until it has finished baking. On a building site, you can’t pour the concrete until the foundation has been dug.
The reason to write it into the plan is what happens when things run late. If digging the foundation takes two days longer, the concrete moves two days later on its own, and so does everything that waits for the concrete.

Start to start: the second task can begin once the first has begun
Neither task has to wait for the other to finish. At a party, once the music starts, the dancing can start. You don’t wait for the song to end before you dance.
On site, once one crew starts digging a trench, another crew can start laying pipes in it just behind them. The two tasks run side by side, and the pipe laying only has to begin after the digging has.

Finish to finish: the second task can’t end before the first ends
Both tasks can be going on at the same time, and only their finishes are tied together. You can start your lecture notes whenever you like, but you can’t finish them until the lecture finishes, because the lecturer might say one more important thing at the end.
In software work, testers check each part of an app as the developers finish it. Testing can’t finish until development has finished, because the last part still needs testing.

Start to finish: the second task keeps going until the first one starts
This is the type people find hardest, and the one you’ll need least often. The second task can’t finish until the first one has started.
Picture a night guard looking after a building. The night guard can’t go home until the morning guard arrives, so the morning shift starting is what lets the night shift finish. If the morning guard is late, the night guard stays on. On a building site, a hired generator supplies power until the mains power is connected. The generator can’t be sent back until the mains are switched on, or the site would be left without power.

Part 2: Timing
Lag adds a wait between two tasks, and lead lets them overlap
A lag is extra time between two linked tasks. When you paint a wall, the first coat has to dry before the second goes on. The drying time is the lag. On site, freshly poured concrete has to harden for about a week before walls are built on it, so the walls get a one-week lag after the pour.

A lead is the opposite: the second task starts a little before the first one has finished. At home, you might lay the table ten minutes before dinner is ready, so the food and the table are ready together. On site, painters can start downstairs two days before the plasterers have finished upstairs. In most planning software a lead is entered as a negative lag, such as -2d for two days.

Part 3: Why they exist
Dependencies are also sorted by why they exist
The four types above describe how two tasks are joined. Project management courses and the PMP exam also group dependencies by the reason behind them:
- Mandatory (hard) dependencies come from the work itself. The roof can’t go on before the walls are up.
- Discretionary (soft) dependencies are choices your team makes. You might decide to finish the kitchen before the bathroom so one crew can do both, though nothing physical forces that order.
- External dependencies rely on someone outside the project, such as a council approving a permit or a supplier delivering windows.
- Internal dependencies sit entirely inside the project and your team controls them.
When a plan needs to be shorter, question the soft dependencies first, because nothing physical holds them in place.
Part 4: In Desktop Plan
How to link tasks in Desktop Plan
Desktop Plan is project planning software for Windows and Mac from the makers of Gantt Excel. In Desktop Plan you link two tasks by dragging from the end of one bar on the Gantt chart onto the other bar, or by selecting tasks and pressing Ctrl+F2. To change the type, right-click the arrow. The lag goes in the Wait box of the task form: type 1w for a week, or -2d for a two-day lead.

Every step, with pictures, is in the user guide chapter Linking tasks (dependencies). You can download Desktop Plan free and try the four types on your own plan.
Questions
Questions people ask about task dependencies
What is the most common type of dependency?
Finish to start. Most links in a typical plan are finish to start, and planning tools use it as the default when you link two tasks.
What is the difference between a predecessor and a successor?
The predecessor is the task that comes first in the link. The successor is the task that waits for it. One task can be the successor of one link and the predecessor of the next.
What is the difference between lag and lead?
Lag adds waiting time between two linked tasks, so the second one starts later. Lead takes time away, so the second task starts before the first has finished and the two overlap.
Can a task have more than one dependency?
Yes. A task often waits for two or more others. Building the walls might wait for the concrete to harden and for the bricks to be delivered. The task starts when the last of them allows it.
Next: the critical path, explained simply
Try the four types in Desktop Plan
Desktop Plan is free to download. Link two tasks, pick the type and add a lag to see how the dates change.
Windows 10 and 11, 64-bit · macOS 13 or later