In geometry, a polygonal chain^{[lower-alpha 1]} is a connected series of line segments. More formally, a polygonal chain ${\displaystyle P}$ is a curve specified by a sequence of points ${\displaystyle (A_{1},A_{2},\dots ,A_{n})}$ called its vertices. The curve itself consists of the line segments connecting the consecutive vertices.

Each segment of a polygonal chain is typically parametrized linearly, using linear interpolation between successive vertices. For the whole chain, two parametrizations are common in practical applications: Each segment of the chain can be assigned a unit interval of the parameter corresponding to the index of the first vertex; alternately, each segment of the chain can be assigned an interval of the parameter corresponding to the length of the segment, so that the parameter corresponds uniformly to arclength along the whole chain.

Every set of at least ${\displaystyle n}$ points contains a polygonal path of at least ${\displaystyle \lfloor {\sqrt {n-1}}\rfloor }$ edges in which all slopes have the same sign. This is a corollary of the Erdős–Szekeres theorem.

Polygonal chains can often be used to approximate more complex curves. In this context, the Ramer–Douglas–Peucker algorithm can be used to find a polygonal chain with few segments that serves as an accurate approximation.^[3][4]

In graph drawing, polygonal chains are often used to represent the edges of graphs, in drawing styles where drawing the edges as straight line segments would cause crossings, edge-vertex collisions, or other undesired features. In this context, it is often desired to draw edges with as few segments and bends as possible, to reduce the visual clutter in the drawing; the problem of minimizing the number of bends is called bend minimization.^[5]

In computer-aided geometric design, smooth curves are often defined by a list of control points, e.g. in defining Bézier curve segments. When connected together, the control points form a polygonal chain called a control polygon.

Polygonal chains are also a fundamental data type in computational geometry. For instance, a point location algorithm of Lee and Preparata operates by decomposing arbitrary planar subdivisions into an ordered sequence of monotone chains, in which a point location query problem may be solved by binary search; this method was later refined to give optimal time bounds for the point location problem.^[6]

With geographic information system, linestrings may represent any linear geometry, and can be described using the well-known text markup as a LineString or MultiLineString.^[7] Linear rings (or LinearRing) are closed and simple polygonal chains used to build polygon geometries.

• Chain (algebraic topology), a formal combination of simplices that in the 1-dimensional case includes polygonal chains
• Composite Bézier curve, a generalization that replaces each straight line of a polygonal chain with a smooth curve.
• Link distance, the number of segments of the shortest chain that links two points within a polygon
• Piecewise regression
• Path (graph theory), an analogous concept in abstract graphs
• Polyhedral terrain, a 3D generalization of a monotone polygonal chain
• Spirangle, a spiral polygonal chain
• Stick number, a knot invariant based on representing a knot as a closed polygonal chain
• Traverse, application in surveying